Not implying 'full replacement' here.. but the 'high cost of transmission' and 'possible zero use of the data in most cases' (since no incident)... Does *NOT* make for a strong case. There are literal behemoth (to name a few; ABC Inc./ Google, Amazon, Meta/Facebook and zillion others) whose businesses routinely have massive intercontinental data transfers. Of-course a physical unit, like you mentioned.. and is commonly known... presently is the only way of ensuring that most data till the *end😪* gets captured in almost all cases. But whenever watching documentaries or coverage.. where they ever said... oh but the CVRs were only designed to have preceding N minutes of data... I often felt a mix of helplessness and being infuriated. Data costs: nope, disagree Cost on environment or impact of server farms, yes perhaps. But (to me) it still feels.. that in our lifetimes.. whether due to regulation or need based 'choice'... fixed wing aircraft... and certainly so helicopters and such.. operating in difficult terrain.. would have it as a 'standard feature'. It = not just 30 seconds - 120 second bursts of data relay... ...but ongoing continual transmission of key parameters. So automated and autonomous to the extent that even some on a suicide mission won't be able to disarmament or disable. Triple, quadruple redundancies... including separate ways to power the electronics or avionics which perhaps sustain for a while even when engines are not working.
@@sailaab Your analogy of the "behemoth" companies that have huge intercontinental data transfers doesn't really apply to airplanes. Those companies typically use either landlines or shipping companies ("Never underestimate the bandwidth of a station wagon full of magnetic tapes hurtling down the highway" -- old joke from the 1970's that's still relevant today) to move data, neither of which are typically available to airborne aircraft. Even when the big companies do use wireless means (satellites) to transfer data, it's one transmitter to one satellite to one receiver. There's a lot of overhead in having a lot of different stations on one network; just ask anyone who's tried to use the Wifi at a busy coffee shop. I'm not saying it's impossible, just _really_ expensive; and since over 99% of black boxes are recovered intact and usable, there's just not a big demand for it.
IF they add a system like Starlink, They can not only send airplane data but 4k video from the copit to their private servers. So they can be used in a case of problems.... CCTV in a plane... but it will be horible to watch the final minutes of the plane and people...
I'm a software engineer by trade. The figure of 20GB stated in the video is probably a reference to *uncompressed* data, and with modern data compression algorithms, can easily be reduced to 2 to 5GB in practice, and can be handled by an LEO Satellite constellation. It may not be necessary for every flight type where accident debris would be left on the ground, but could be accomodated for transoceanic flights, like MH370. Think of money spent for months of fruitless ocean searches to find that plane, and the cost isn't nearly as onerous. The number of parameters sent such as simple latitude, longitude, altitude could make recovery of the physical black boxes a possibility as well, and should also be less expensive. In short, there is no reason other than corporate inertia to not attempt to stream "black box" data to satellite, then the cloud.
Yeah, if you can literally store data on a disk you are largely unconstrained about what you can record and store. But if you look at something like video game networking, they really just transfer inputs across the wire with each game client replaying a simulation which is not even a kb/s of data. You can downsamble, compress, and prioritize to bring this down to virtually nothing. There's also only about 100,000 commercial flights per day in the world which is an absolutely trivial amount of data to handle.
Absolutely. We're not even talking here about needing a perfect record of all inputs and voice data. There are many ways to truncate the data so streaming/storage is manageable *before* we have to consider data compression. As you say prioritising data for high-risk over-water flights might be a good start, and then we might choose to prioritise information that allowed us to find a missing black box. Even constant telemetry data could be trimmed to prioritised controls. Once a flight lands, there isn't even any need for storage. This is far from an intractable problem. My feeling is that the real reason is that the *initial* cost of developing a robust system would be significant and no company really wants to pony up the initial outlay.
Also a software engineer by trade - compression works great when you can leverage the patterns in the adjacent data to de-duplicate stuff and save space, but its a LOT harder when you are talking about real-time data. I don't even mean real-time like a movie video decoder, but real-time like aviation and battlefield systems process that are so tight a loop you can't even use a normal OS but instead use customized stuff like RedHawk and VXWorks that can ensure there's zero delay in processing time. Its also possible the 20GB figure might already be compressed (he doesn't say in the video, and we don't know for sure). I've not found any examples in my job where real-time data processing is compressed on-the-wire, even if its compressed at-rest.
exactly. the idea of keeping your only backup of data on-premises would make any IT professional cringe. the amounts of data mentioned are frankly trivial in terms of modern bandwidth and storage capabilities.
I worked at Satelliite and Wireless company for many years, and in 2001 was on a project to provide data links to aircraft. After the Sept 11 attack, we proposed that the data link also carry camera information. We did a video conference call with a Congressional Committee from the test aircraft to show viability. They said "great idea, now who's gonna pay for it"....and there it ended ....
"Most black boxes are able to be recovered so why bother backing up the contents?" Most airplanes don't crash, and yet when they do we take steps to avoid them in the future anyway. The entire ethos of the aviation industry is that the 'same' crash should never happen more than once. The fact that there's been more than one plane whose black boxes couldn't be recovered should be viewed as a failure of the industry. "It's expensive :((((" I'm sure all the million other safety features on a plane are expensive too, but they're all still there anyway. That's why we have aviation safety regulators like the FAA - to prevent people from just cheaping out on safety equipment.
Just think how many crashes may even be able to be prevented if the black boxes could transmit in real time during an emergency to the ground and they can possibly tell the crew where the problem is and maybe solve the problem enough to be able to land the plane. That could save lots of lives.
@@KevinBenecke I think it takes the ground to long to answer, if it is important. Else the plane itself would warn you/send an error code. If you have that time, then you could also have easier way for the most parts. ("Land" on a airport, with the firefighters ready and so on)
Imagine if the top 20 richest would die in plane crashes. I'm sure $20 billion to protect them all the way to the ground would still be "pocket change" anyway then.
@@KevinBenecke While reasonable on paper, I think something like that could be counterproductive. From this channel, I've gathered that in the event of an emergency, air control keeps OFF the radio as much as possible with the plane in distress because they know the pilots are extremely stressed and focusing all their efforts on getting the plane back under control. Interruptions are kept to a minimum.
I think that they should keep recording data locally on the aircraft as a backup when or if they start streaming the data. That way (if the device's data is recoverable), it can fill in data gaps from loss of live communication post-incident.
The thing is if they can start streaming the data when an emergency happens, they may even be able to prevent the crash from even happening. Ground control could possibly discover the problem while in flight and direct the crew on how to resolve the problem enough to be able to get to an airport and land the plain. The transmitting doesn't have to begin until there is an emergency. There should be a button in the cockpit the pilot can press to start the streaming. Plus there should be a system that automatically starts the streaming if a problem is detected which could help to prevent the problem from even happening because ground control could radio to plane and let them know there is a problem happening to which they can get to the problem or land the plane before the problem gets worse. That could save lots of lives instead of waiting until after the crash happened. A lot of the crashes that have happened most likely could have been prevented if they could do this.
the pilots already have alarms and an entire dash of performance gauges in the cockpit for them to know a problem is happening... what you're really asking for here is a check-engine light so ground engineers can watch your 80+ parameters remotely on a screen, never look at or inspect anything, and reliably find a solution based entirely on that - in real-time... don't think that's going to work out
@KevinBenecke True, an ai could be monitoring the data stream and bring up an alert to a human (eg ATC) when it notices something starting to go wrong.
Agreed. No reason in today's day and age not to just have both for redundancy. In fact I'm surprised it isn't mandatory for intercontinental flights. Gaps in the data? Find the box. If no gaps? Well no need to spend god knows how much to go searching in the ocean for an incredibly tiny piece of wreckage that could be anywhere in a huge radius. Seriously it can't be that hard. Ya know those planes are streaming more internet.
While it is fairly infrequent that the black box is missing or too damaged, in an awful lot of your accident report videos the pilots forgot to pull the voice recorder breaker so we don't have what was actually said for the investigation. This could also be easily fixed by having about 1tb of solid state flash in the black box, to allow for months of recording to be stored, and would also be fixed by having a 700kB connection to a satellite (that a lot of planes already have anyway) and storing it off site
this is what i dont get. if there is any reason why the blackbox cant store so much data, just save the data secondary normally on any hdd/ssd inside the plane. The data is lost in an emergency situation, but atleast in most other situations the data can be saved and recovered
But where you are directly transmitting to, from the perspective of the aircraft, may often be a satellite instead of a ground antenna, where it goes from there is less of an issue.
@@subbss it wouldn't be an issue unless the aircraft diving towards ground. Then it can't transmit any data to the satellites anymore. For further arguments, there's already been discussion in the comment sections for the why's.
To me one major benefit to streaming this data (in addition to the black boxes) is that it could enable more long term storage of this data. In a lot of your crash Investigation videos there was critical information missing because it was more than 2 hours old. With long term storage you could look at older information even from different flights which might help draw better conclusions. Additionally, there is a huge potential for algorithmic data analysis to predict failures before they occur. I'm assuming the airplanes have their own diagnostic systems already, but in industrial settings long term data is incredibly valuable as you can pick up on very minor changes and use it to prevent costly machine stops.
@@philipwhiuk it really isn't. Data storage is complicated when it's still supposed to work after being pulverized along with the plane... In fact the very first example mentioned showcases this.
You are totally right: it will be possible to have supercomputers running an AI to detect potential failures on the ground. It is just a matter of time: sooner or later the costs will drop down and for the companies, the savings in maintenance and reparations will overcome the cost of transmitting data..
Flight data is analysed for maintenance purposes and is important part of the reliability. The amount of data recorded is only going up as we see the benefit.
It was not the outside of the boxes that had to be painted black when they were using photographic film, it was the inside of the box that was painted black, to absorb any stray light. Black box was a term that originated in the theatre/film industry to refer to a stage or set that was painted totally black to absorb any stray light. The term was later applied to any container painted black inside to absorb light and protect film. As the early flight data recorders were using photographic film, they were painted black inside and hence were referred to as black boxes.
Most of the settings will be relatively easy to store - you only store deltas. Whenever a switch is changed, you get a 3-byte (or thereabouts) change. Most switches are toggled less than 10 times per flight, so all of those fit within 3 kilobyte - over the full 20 hour timeframe. Those that measure a constant signal will record whenever they update - which is typically a max of 60x per second, with an accuracy of at most 32 bits. They are also typically ideal for Huffman coding - flight height does not change immensely quickly very often, and nearly all deltas will be very close to zero. Then the audio itself. If you assume the audio channels will be compressed to 384kbit/s (assuming 6 channels at DVD equivalent quality) you should send a packet of ~50kbyte per second to keep up, of which 96%+ is audio. Using variable bitrates should bring that down to ~20kbyte per second given that most of the time, most of the channels are effectively silent. That becomes 72MB per hour - very very doable with satellite, but still too expensive to actually do.
Only upload data when the avionics detect an anomaly. 30 minutes of data = 36MB on a small percentage of flights = quite affordable especially with how cheap it's getting now. Also a switch change only has to be 1/8 of a byte unless you're monitoring the analog voltage to see deeper into the wiring, slight arcing, etc.
@@thorlancaster5641 the con is, how would you know what were the pilots reactions and all stuff in the case of a sudden depressurization caused for example, by a bomb ?
@@User-jr7vf you wouldn’t, but you wouldn’t know that with a modern black box system if the bomb destroyed it either. And unlike a black box you wouldn’t guarantee it’s loss if a plane crashed in the ocean. Which is a much more likely scenario then a bombing.
I worked on the Navy P3 Orion aircraft in the 1970's. It had a flight recorder mounted in the tail which will be ejected in the event of a crash. How do I know this? I accidently ejected one while I was doing some work in the cockpit when I bumped the switch. It successfully ejected and landed on the ramp. Of course my supervisor gave me the task of reinstalling it in the tail. It was an embarrassing event but memorable. The P3 also has "crash switches" in the nose that will trigger the eject mechanism automatically during a crash. The recorder is mounted in a fiberglass airfoil that can "fly" to the ground and it also floats. So this technology has already been around for awhile.
Gary, typically, black boxes are crazy heavy. Are the ejectable type lightweight? If so, can they survive staying on board a crashed aircraft if the ejection system fails? If aircraft power has failed (except, say, the battery), can it still eject? On many civil aircraft, the recorders even stop working when main bus power has been lost.
@@ImperrfectStranger This recorder was not super heavy and was mounted inside the fiberglass airfoil. I don't know if it would survive without ejection. I suppose it depends on where the tail ended up at and if it was exposed to a post-crash fire. But with the crash detection switches being in the nose, that's usually the first point of impact so chances are good it would be ejected before a power failure. Unless of course there was a power failure before the crash.
@@ImperrfectStranger Hence the pitiful state of accident investigation data. Many times in the past accident investigation was delayed and hampered by the fact that the recorders simply failed due to lack of maintenance. This has been improved by digital recording, removing the delicate moving parts from the system, but really, how hard would it be to add extra capacitors to allow recording to continue, even for a few seconds after power is lost? Another item that should be added is a non-silenceable alarm that is triggered by removing power from the recorders, and instantly turning on the emergency locator beacon(s) of the aircraft if this happens. That would prevent "accidental" failures of the black box circuit breakers. I will leave it to the reader to investigate why none of this has been accomplished on passenger and freight aircraft.
Doesn't ejecting the Black box before the impact of the plane makes us lose some data? Not even sure if the Flight Data of the "tumbling" is useful, but it sure does look interesting in some documentations.
Aireon is already doing this. In fact, the FAA gets real-time data from the 737 max fleet via the Aireon network to monitor for issues, although the data points aren't nearly as comprehensive as the black box. Starlink does not capture adsb. However Aireon is deployed on the fleet of next-generation iridium satellites.
Starlink does not capture adsb right now yes, however the aircraft could have software installed to send the data out over it's Starlink link for Starlink equipped aircraft.
Folks, SDR doesn't change the antennas or amplifiers. Starlink has Ka, Ku, and E band support. Adsb uses 1090mhz. Won't work without the right antenna and RF front end. SDR means protocols can change but doesn't mean software defined physics.
Aerospace software/data/cloud engineer here. ADS-B is all very well unless the cockpit crew disable it (as appears to have happened with MH370) due to a rogue pilot, pilot error or hijacker duress. Also, the cited data volume assumes no compression. Cockpit data are VERY compressible, potentially down to about 1% of raw data volume, which makes the costs of routine transmission and storage fall into the noise. Those data are an absolute gold mine. By comparing similar flights on the same airframe over time, it may be possible to identify and predict failure. By analysing over a whole fleet, even over all airframes of a given type, there are opportunities to improve everything from design to training. You can even look at accelerometer data over time, correlated with GPS, to get fine-grained analysis of meteorological trends. There is a saying in cloud technologies and data analysis: there is no such thing as too many data, only the wrong type. The potential for streamed aircraft data goes so far beyond crash investigation that that becomes almost a sideshow.
It would be handy to have this data stored on a server. In the event of a crash, investigators would be able to immediately access vital information while the other investigators go to the site.
It wouldn't be done continuously. Aircraft have have had the option of transmitting "snapshots" of (limited) data for many, many years.... either by manual command, or automatically at certain points during the flight. Larger, bulk transmissions are often done in the vicinity of airports where VHF groundstations are used, rather than satellites.
I was thinking of RR and their Trent series. I seem to recall though that they have a dedicated team constantly monitoring the system and it's paid for by the aircraft owners who lease the engines. Thinking of the comment Petter made, the airlines would pass this onto the public through ticketing prices and it wouldn't sit well with the budget companies.
I’m a long term satellite comms professional - worked ten years for the leading provider of aircraft satcoms systems. The cost per Gb is not really a problem - $10/Gb for the satellite bandwidth. There’s a big cost in fitting the antennas and modems, maybe $1m if it’s not a line fit (ie built into the plane by Airbus or Boeing). Also while reliability is good, a typical satcom system would not provide data in a dive situation like the China Eastern crash as the ‘look angle’ of the antenna can only cope with a limited range of attitudes. Certainly it could be a useful addition though, especially where aircraft are being fitted with broadband satcoms for customer internet access, in which case the cost for the extra bandwidth needed for CVR and FDR would be marginal.
@@darrinito I am assuming the new method will have less antenna gain? I suppose being at 35000 ft means less atmospheric signal loss though so it would still work ok.
@@darrinito Ok thanks for the information. I work with radio telemetry systems but we don't use antenna systems as you describe. Off to Wikipedia to see if I can understand how they work. Thanks for the information.
@@geoffwoodgate7450 The key factor is the frequency band in use - Inmarsat and others provide service at L-band (~1.5GHz) and don’t require accurately steered antennas, but the bandwidth is low - too low for this type of application. For broadband connectivity we use Ku or Ka band (12Ghz, 29GHz) and at these frequencies you need quite a large aperture to get gain, plus ability to track the satellite to ~1 degree accuracy. This is tricky when the aircraft banks or dives as the radome housing the antenna on the top of the fuselage is quite low, restricting the antenna movements (whether mechanical or electronically steered). Look at the ‘hump’ on an aircraft and compare with what you see on a ship that doesn’t have aerodynamic constraints.
You're right in that they'll become supplemental system and not a replacement system. The broadcast system is going to be more succeptable to issues than the hard copy on the plane. Any extra data is welcome but they won't replace the "black box" system, just enhance it.
Fun Fact: Marine industry is also using “black boxes” named Voyage Data Recorders. Systems installed after the 1st of June 2014 are mandatory to have two different final recording media (that keep the same data). One fixed with a beacon and one that deploys automatically, called float free that starts transmitting it’s position at 406Mhz.
Another possible source for the name "Black Box": In engineering terms, a black box is a device or object we insert into a design per specification where we don't care about the inner details. The first BB's inserted into a plane was just like that: Mount this box and hook up those cables. It doesn't contribute to the characteristics of the plane, so don't bother what it does.
Black Box also often refers to a device or programming function where we know what it's supposed to output but have no idea how it does that. The inner workings are hidden from the user.
4:15 in my opinion the term black box is a carry over from normal science/engineering usage where any piece of equipment that we don't need to understand in detail is referred to as a black box. A pilot and and almost any plane designer don't need to understand how the insides of the recorder works: so to them it's just a black box. The only thing a pilot has to know is how to stop it overwriting the records immediately after a survivable but serious incident. Even that is not a feature of the black box itself: typically the pilot would simply pull the circuit breakers on landing in those cases I'm old enough to remember a crash report on the TV children's programming from sometime in the sixties or seventies where the black box was shown being retrieved. The TV presenter explained that it was neither black nor a box. It was roughly a cylinder but with rounded ends. The shape made it more resilient in a crash and the bright orange colour (as he described it to viewers like me watching in black and white) made it easier to find. He suggested we should perhaps call it a red egg, but I don't think anyone in aviation ever used that term
Nope, the term was developed in parallel--if anything, the systems engineering usage was influenced by the aviation usage, but there's scant evidence for that either. As Petter said, early data recorders were usually secret and secret components were often painted black. Literally a black box. In circuit diagrams in the early 1940s, a heavy black box was used when a complex component wasn't relevant to the operation of an enclosing circuit. The diagramming conventions for this eventually changed, but the term stuck. It's possible this was influenced by the habit of painting secret gadgets black, but little evidence either way.
@@herestoyoudoc Vitold Belevitch puts the concept of black-boxes even earlier, attributing the explicit use of two-port networks as black boxes to Franz Breisig in 1921 and argues that 2-terminal components were implicitly treated as black-boxes before that. Quoted from en.m.wikipedia.org/wiki/Black_box (Scroll to history section) Avionics can be said to have started with IFF during WW2, or arguably when radio transceivers started to be used in the air, both of which came after 1921, of course. It didn't become regarded as a subject separate from general electronics till the fifties or maybe even the sixties. I stick with my original assertion that aviation borrowed a term that was already well established in scientific and engineering usage
@@trueriver1950 I'm not saying that it's impossible for the etymology to be the same, but the evidence is really scant and speculative. The simpler and high-probability explanation is that data/voice recorders were originally literally in boxes painted black and the name stuck. Or, ultimately, with the additional background offered from Breisig, it could be the etymology is shared but by analogy to a camera and not for any other reason.
The US Navy P-3B had a data recorder that was installed on the tail. In an accident, the recorder detached from the aircraft and activated an onboard transponder. The data recorder was also designed to float if the accident occurred over water.
Loving the title: straight to the point telling us what to expect.. youtube nowadays is oversaturated with titles in the line of "why THIS plane failed" etc and it makes me mad. Your video was an insta-click because of that in combination with an interesting topic rather than a fishy clickbait.. much appreciated.
Given the number of incidents you have covered where the problem originated on a previous flight, how about downloading the data during turnover at the terminal? While it may not be helpful in all cases, it could still be helpful in some. This would also allow investigation teams to access even older data to see if anything was missed during previous flights.
My thoughts exactly, although 88 flight instruments' data could easily and cheaply be streamed live to detect a potential risk live on the on-going flight as well, not just future flights. But yeah, we now have ginormous AI machines that would just love to swallow up all this data and detect patterns across airplane models or flight conditions, maybe even specific pilots, if this information is used positively and productively (training) rather than retaliatory.
The black boxes should always be in the planes as a last resort fail safe option even if they do get this up an going at a high capacity. Redundancy is the key with planes and having more than one way of keeping track of all the data is going to play in the favor of them and the airlines and especially the people that rely on them every day.
Trey N said, "The black boxes should always be in the planes as a last resort fail safe option..." ------------------ You make the point well-- too many critical systems succumb to a single point of failure. Likewise, a broadcast stream of data always benefits from another format-- an onboard recorder. Post-crash data is usually bought at great price, and having an alternate system is a cost-effective remedy.
another possible benefit from contiuously uploading to the cloud would be that someone at corporate could look and find and suggest solutions to issues.
I'd love to know why the regulations still only require 2 hours of cockpit voice recordings. Especially when ocean going cargo ships keep the entire weeks long trip
I'm not sure how often this happens, but I've seen a couple of cases where the black boxes lost power and stopped recording early. They should hand an independent battery backup. It would be minimal expense and ensure up to the last second data that could be significant in stopping future incidents.
essentially all that would be needed to be transferred to sattelites would be it's position, because in that case they would be able to recover it's location so we don't have unsolved mysteries like MH370
@@MentourNow Indeed and with satellite based ADS-B receivers we get global coverage. What it won’t stop though is a rogue pilot pulling the breaker on the transponder, or on a satellite internet system before deliberately losing the aircraft, as may have happened in China Eastern and MH370. The flight crew always need that option in case of fire or malfunction in the electronics
This, exactly this! Everyone keeps saying it's too much data for every plane to contanstly transmit black box information. Commercial flights that have 100+ seats should be made to transmit latitude and longitude every 2 minutes. Is that really too much data to deal with? Make the system on by default, you can't switch it off. It'll be much eaiser to find black boxes.
imo, there's a lot of stuff that could be done with compression, and intermediate data packages in case of value spikes while maintaining a lower update rate outside that for regular purposes. for a "not a replacement system", but a complimentary system that provides at least some data for these situations where the data recorders are not recovered, or damaged. especially combined with panic buttons that both software and pilots can trigger, to start transmitting more detailed data, this seems like a very good addition to existing systems. it could also provide ground crews with additional tools to support aircraft that have not yet crashed, but are experiencing problems - hopefully preventing the crash all together, by allowing a team of engineers on the ground to double-check the status of the aircraft systems and helping the flight crew troubleshoot the aircraft. who could even have access to a whole different level of tools to look at the data - like, if there's a suspicion of a problem with a specific engine, they could with just a few clicks see the performance graphs of that engine during the entire flight, and even get comparison data of the previous flights to form a hypothesis of what's going on and what steps could be taken. like "yeah, i can see that the number two engine has been running x degrees hotter than the number 1 engine, and has been holding that level steady for the last y minutes. there's also been a drop of hydraulic pressure back then, but not since" it could also allow for a more in-depth tracking of the time some systems have been active, possibly leading to improved maintenance intervals - like, on this aircraft, the flaps have been deployed for this many minutes at speeds between x and y, exceeding the "normal" rate of their deployment at these speeds by some margin, and increasing the load on the involved systems, so they should actually be checked a bit earlier than the regular maintenance interval. it'd also allow for some external validity checks that might not be feasible to implement on each aircraft - like, for example, it could gather the metar data from all airports in real time, and check that aircraft at or below certain altitudes near the monitored airports have an altimeter setting that's within some margin of the actual current meteorological conditions, and if not it could alert a person on the ground to the possible mismatch, providing the live data from aircraft and ground plus the latest metar update, plus maybe information from other nearby aircraft for comparison, and that person then can determine if they might need to contact the aircraft to alert the crew to a problem. i'd imagine that there's a lot of checks that could be run on the ground side that should get filtered through a trained human before alerting/distracting the pilots in the air. things that can be easily expanded upon without requiring updates to all kinds of onboard systems, or being flight certified, since they don't run on an aircraft system and get double-checked in a non-critical environment before being brought to the attention of the pilots.
A lot of your (good) ideas are already implemented as optimizing of the maintanance and in result avoiding unplanned availability of aircrafts saves airlines a lot of money. Your idea of having a person on the ground who is looking for possible mismatch in the data of an aircraft would be very expensive as it adds a lot of headcount. I agree that a lot of checks could be done - but soon as you aware of that should be checked to identifiy possible mismatches it is easier to let a computer do all these checkings. A lot of accidents are caused by wrong sensor data confusing the pilots which data they should trust on which is understandable in a stress situation where you get mutually contradicting data. I think here additional software could help the pilots a lot by telling them how reliable each data source is. For example if one altimeter tells you (by defect) that your are flying at 30000 feet while the other tells you that you are on 3000 feet the software could use additional data from GPS to determine which of the both sensors most probably is wrong. Even the outside temperate could help in this decision - at 30000 feet is is always much colder than on 3000 feet. Engineers can spend a lot of time to think about and define such realiabilty calculations for each sensor using data from other sensors and a software on board could use these calculations even in the most stressful situations to show them which instruments they should trust and which not in case they are in doubt.
@@elkeospert9188 thanks for your reply! > "I agree that a lot of checks could be done - but soon as you aware of that should be checked to identifiy possible mismatches it is easier to let a computer do all these checkings." i meant that, you have computers run those checks automatically, but do not automatically send information back to the pilots. rather, the information that gets flagged by the computers then gets double-checked by qualified people (who only check those flagged data points and not monitor everything all the time) before the pilots get notified. the pilots might be in a very stressful situation, and i'd rather not have automated systems on the ground add to their workload, but have a trained human being assess the situation and make a judgement call there > "A lot of your (good) ideas are already implemented" glad to hear that they've had thought about that stuff too... though i'm not really surprised, tons of good people working on improving those things :)
I have to question your data volume assumptions. 88 parameters using 2 bytes per sample of data updating 10 times per second would be 14 kbps of bandwidth. That's old timey modem speeds. Taking audio from 3 sources, moderately compressed, it would add around 480 kbps, totaling 500 kbps. That is such a small amount of data, when these airplanes already offer internet services in flight. By all means, keep the black box, you should always have a physical backup you could access in case something goes wrong. But, you'd save a lot of cost and time if you could always access the black box data (even in a limited form) instantly after an incident, and never have to worry about spending the time and money on a search. Sure it is rare for an incident where the black box cannot be recovered, but not at all rare for an incident where it is badly damaged, or cannot be immediately found.
I also mentioned this point in my comment. In its principle I agree with you - it’s not a lot of data. I would go on and argue that it’s actually much lower than 500kbps. Your calculation is correct when the audio is analog and uncompressed. But a digitalized and compressed audio file (which makes it worth transferring over data) can go as low as ~10kbps without losing much quality in scenarios like a cockpit.
@@Marcus-cc72 how efficient modern codecs are is insane. If all you need is enough quality to say hear the pilots speaking you could go as low as 6kbps or perhaps even less and it still be understandable. (You'd lose a lot of quality in things like background noise which of course could be important in an investigation, but as a backup method if the blackbox can't be found, it would be fine). I'd guess with compression you could constantly broadcast a handful of key parameters and voice data with the same or less amount of bandwidth as a 56k modem.
My suggestion for numbers. 128 parameters at 16 bit resolution, 1 sample per second = 2 kbps . Voice only needs telephone quality - 3.1 kHz bandwidth. Min sample = 2*bw = 6.2 kbaud At 8 bit quality = 49.6 kbps per channel. Allow 8 sound channels = 400 kbps We're all getting the job done with 0.5 Mbps data rate - 1/40,000th of what our host suggested.
@@elijahwatson8119 right with you here. I’m not the best informed in terms of sound file encoding but I’d expect it to be around that ballpark too. Plus CVR quality is already pretty nasty anyway.
I worked on a prototype study back in 1998 which looked at the feasibility of doing live data transfer for the black boxes. At the time I was calculated that this would only become economic when the cost of transfer fell to below $1 per gigabyte of data. I think we will probably be there in about five years.
@@ChrisD__ At the time it was about $20 a gigabyte. Not sure what it is now as I am no longer in the industry. The last price I heard, which was in 2010 was $8 a gigabyte.
@@mortgageapprovals8933 You are dealing with a lot more than GPS data, which is fairly small. On the study I was involved with we had to record over 7000 data points, each of which was a 64-bit data value. The requirement was that we sent the data point identification, the time of the data read, the 64-bit data value and a checksum that is 18 bytes of data per data point or 144 bits. Each data set had to be read every 10th of a second. Add to that the cockpit voice recorder data get a lot of data having to be upstreamed. With flight data about 4.5 gigabytes per hour and cockpit and cabin voice recording coming in at about 1 gigabyte an hour (multiple pickups being recorded) the test ended up running at just under 6 gigabytes per hour.
We can never replace the blackbox but i think, while having the blackbox i think sending extra copy to the server during abnormalities would help if the blackbox doesn't survive, even the litle data sent out before a crash can give an idea closer to solving a tragedy, would give a general idea.
Why not just send the voice/data recorder data through ground/satellite with a lower resolution? (e.g. 128kbps audio quality, and less frequent parameter sampling, which would require a data transfer rate of less than 1Mbps). I'm not suggesting this is as a replacement to the 2 recorders, but as a complementary measure. The cost of this surely won't be prohibitive (both to implement and to pay for data transfer). This could certainly help in those rare cases like China Eastern 5735 and MH370, and is better than nothing.
Somebody else posted in another comment that it wouldn't help. In a dive like 5735, the plane to satelite connection would be lost because the transmitter/receiver would no longer be pointing at the satelite, simply due to the orientation of the plane.
The broadcasting cost is one issue. The installation/design/certification cost is another. You still have to justify the cost for rare suicide events. Not mentioned in the video are the human issues. Not all pilots like "big brother" listening to them. I don't know if this is still the case, but pilots used to have the option of deleting the audio at the end of a flight (when the park brake had been set).
@@danielch6662 Services like starlink use phased array antennas (meaning the "beam" is steered without any physical movement), and if correctly configured and with sufficient satellite coverage, should still be able to communicate with satellites even when vertical.
About uploading data to the cloud: Sending data after declaring pan pan or mayday... (Or punching in some code by cabin crew) would reduce the amount of data sent under normal circumstances, but in an actual emergency scenario it should be helpful. If it is sent in a form that enables training pilots, engineers. Using high amount of bandwidth and storing a lot of data ONLY when needed would be still affordable. Even if we would allow the aircraft to initiate this data transfer on its own, in reacting to TCAS and EGPWS warning, or maybe in reaction to some fault, that would enable airlines and other operators to start reacting to emergencies before the pilots even have a chance to make a mayday call. Of course these networks can be made in a way that they would allow 2 way communication allowing to help the pilots with some visual information. I would prefer if the data broadcasted would enable people to see the developing emergency in a simulator. If simulator training staff, who should routinely deal with various (simulated emergencies) can quickly understand the developing scenario and can communicate with the rest of the company to help the pilots as needed, then this communication system would do MUCH more than replacing the black box.
Replicating an emergency situation in a simulator on the ground, is a great idea! The fresh mind of the simulator pilots would certainly help where the aircraft pilots are mesmerised by a stuck autotrottle, or being confronted with the 737max nose dive, where it was necessary to take the unnatural decision of decreasing the speed of a falling aircraft in order to regain authority on the control surfaces in manual mode. Well said.
Except in the vast oceanic expanses, there isnt enough data to send a text - let alone what you suggest. Unfortunately, too many people carrying a smart phone think the rest of the planet has the same coverage - it does't and its not even remotely close.
@@grahamo22 Except for your comment with the implied insults forgot at least 4 different problems. First of all, your GSM phone is a land based technology, so of course you can't send anything where there are no towers... But it has nothing to do with data or bandwidth, but with coverage of a single technology But there are global satellite based communication systems, and even a few decades ago they were used even for live TV. Your lack of knowledge and technical understanding wouldn't be bad if you wouldn't throw implied insults only backed up by your lack of knowledge. Second: Existence of both long range and satellite based communication were acknowledged on the channel, when you pretend that they don't exist to throw implied insults... That is already a bad idea. Anyone interested in civil aviation would see that people often review on board WIFI and it is easy to be aware of the feasibility of some internet connectivity that can be shared by a lot of devices. So your offensive comment didn't only ignore old information about different networks of technology, but also common knowledge about aviation and on board services. Third: The video explained that it is easy to find the black boxes too because most emergencies around airports, also shorter flights over land have worse statistics, etc. so your argument about oceans is totally irrelevant in majority of cases. A dick comment that implies that others are dumb and ignorant, while shows only your ignorance is usually not a good way to engage in a civilized conversation. Fourth: We see more and more satellites for different purposes each and every year. Even if we ignore existing technologies mentioned, that wouldn't prevent deploying new satellites. ADS-B Satellites are also somewhat new and mentioned in the video, and new satellites are often made for new purposes. So while you might be unaware of things like Inmarsat: www.inmarsat.com/ And you probably don't know that the Satellite Data Unit of Malaysia Airlines Flight 370 was in contact with Perth, how they used the "Classic Areo" plan at the provider (yes, the name implies, newer and better plans) and that Inmarsat S EAN is EAN for "European Aviation Network" Or you don't know Globalstar either, as they aren't that well known services, Iridium was in the news often, etc... But these companies aren't only used by aviation or news industry, but they often provide a communication architecture for ships also and can help with providing internet for tourists on cruise ships. The need and existence of these systems are somewhat common knowledge, but even if you don't know about the existence of these systems, you don't know the global satellite based internet service provided by Musk (and used in Ukraine by their army too) so you managed to ignore all the news and common knowledge, and you lack the upbringing to know that you shouldn't paint others as ignorant when you lack the basic knowledge about this field, it would be still common sense to expect development of (more, better) satellite based communication services.
@@grahamo22 you can connect via satellite ANYWHERE in the world, but right now it is expensive for commercial use... How do you think a plane can provide wifi connection to its passengers if not connected to a satellite?
I don't think it would ever be reason enough to get rid of flight/cockpit data recorders - even if we got to a place where it was possible to stream all the data reliably in real-time. Why would anyone ever get rid of that belt and braces backup? I like the idea of the bursts of data if something feels off or that the pilot can manually send, and that it records/sends x amount of data before that point. It reminds me of back in the day when dashcams only saved the previous 30-60 seconds of footage before a crash or whatnot. It seems like a really good compromise considering the current safety and data transferability.
Redundancy is the name of the game when it comes to airlines, so I think that the data burst idea in addition to the physical black boxes is an excellent one. There is always a chance that the black boxes will not be readable thanks to damage, and there is always a chance that the data will be corrupted, so they can cover for each other.
You can compress data, also you can also prioritize various data streams from instruments. So it doesn’t have to be as high resolution as whats typical in the actual black box recording
@@TheExileFox I make systems that tolerate intermittent connectivity all the time. It sends data, gets an acknowledgment of what was received. Keeps trying on the segments that didn’t make it through. You would still have a high resolution recording on the planes black box. What’s uploaded to the cloud would be something skeletal in comparison
@@NetAndyCz those are included in the category. for most people a computer is black box technology. for some, just about everything is. addendum: it is also applied to stuff with no user serviceable parts inside.
Mentour- my question is why are there no cameras on airplanes? Today cars have them to reverse, to aid with side view blind spots etc. When you go into any number of public places there are cameras watching your every move for security reasons like banks, casinos, hotels, etc. So many issues even after finding the black boxes leave questions unanswered as to what "really" happened. Cameras should be on board in almost every spot possible inside and facing outside of the plane. If a captain needs to check if a engine is actually on fire or leaking fuel or is landing gear locked/down he/she should be able to flick a switch and see it on camera. If a passenger is causing an issue a captain or even in the flight attendant galley should be able to see it on a monitor at any time. All this should record to the black box as well. The cost of this technology is negligent and the size of cameras now would be invisible to the public but so helpful in every way from day to day use, safety and trouble shooting after the fact in case of accident. Mr airplane executive, we can offer you our new A-320 plane standard features at 100million$ or do you want the upgraded version with cameras/monitoring with upgraded blackbox for 20000$ more? No brainer to implement!
Many countries have privacy rules that prevent recording of employees or customers without justifying the need. A brief, voice only recording accessed only AFTER an accident is one thing, but the legal quagmire of multiple national laws involved, for the extremely limited potential benefit, make the idea of random video surveillance something that probably just isn't worth it.
Mark is right - and it's not just countries. The GDPR is quite simple in this: unless you have a very good reason, any form of camera placement is absolutely prohibited. As an example: if you have a supermarket and you want to put up a camera in the stockroom, you can only do that if you can show that products are disappearing. Besides, how would you as a passenger feel if your suggestion was implemented - including the lavatories?
could implement a camera that start recording visual and voice commands whenever 7700 is squawked, which could be compressed even streamed, so that if for any instance all options appear to be have been exhausted , more Professionals on the ground could offer advice in real time.
I really think this is a great idea. I don't see any privacy issues because the recordings are deleted as soon as the flight is completed. Cell phone cameras are now so small, so good, and use so little power at the cost of about $20 or so. This would have been very valuable in the 9-11 hijackings and the china crash. Like the cockpit voice recorder, the actual footage would never be released. Only a description of the events. I'm pretty sure the NTSB has been pushing for this for many years and it's being fought by the pilots union.
For me as a computer scientist this was a question I had for quite a long time. Sure, live streaming of data isn't a good solution for the reasons you mentioned but that idea of sending data only event-related sounds like something that should at least technically be absolutely possible. And with the costs going down quite fast, it'll be not long when it's be cheap enough to maybe be enforced by authorities one day.
20 GB seems quite a lot. If 100 parameters are sampled as single precision floats at 1 kHz that would be 1.44 GB/hour. And another 1.44 GB/80 min for 4 channels of uncompressed reasonably high quality audio. (2 CDs)
@Foobar You forget that it's probably not only random float values thrown into the blackbox (or into the streaming to stay in that idea) but rather raw flight data along with tons of metadata and some stuff we as amateurs don't see. So you cannot take e.g. valve position as one raw float value but you have to consider stuff like "which valve is it", "actual flow rate", "normal flow rate at that valve position and at this phase in flight" and maybe much more. And all that needs to be packed into a processable format which in itself also produces some data overhead. And this is not considering networking and protocol overheads and so on when you do live streaming of flight data. In the end all that may be absolutely possible technically. But as long as lost blackboxes are that rare and live streaming blackbox data isn't mandatory, it comes down to cost and value. And even if safety is (or should be) number one priority, it's not a big secret that airlines try to safe money whenever possible as long as it's not impeding safety.
@@Sinned1208 Yes, but it doesn't need to repeat the metadata millions of times, that would be an even more terribly inefficient format. There just has to be a lot more than just the 88 flight parameters for there to be 20 GB of flight data.
In order to receive good amounts of valid data, you need bigger bitrate for reliability and stability. Ping also plays an important role in data streaming. Most of us tech junkies relied on fiber optic and transoceanic internet connections in order to get stable service while airplanes are completely wireless while airborne . And trust me, live broadcast TV expend hundreds of gigabytes of bandwidth everyday just to stream video and audio. Flight data is way too complex for another computer to compress it then transmit it.
While black (or orange :p) boxes won't be eliminated anytime soon, it would be nice to keep improving them as technology is advancing: make them record more data (how about 24h of CVR for example), make them more durable, easier to find, and easier to handle after recovery.
If airlines start using Starlink for their passenger data anyway, could they potentially use a data channel on it for flight data? Maybe not the full amount that the black box would record (especially not the voice recordings), but basic telemetry and errors/warnings maybe? Might be a fun project for your discord to chat about what would and wouldn't be useful to send that way
Well, at the (not too distant) point where Starlink will be providing fast internet access for everyone on a plane (i.e. half of the passengers being able to stream videos at 1080p or more), the 20GB of telemetry and voice on a long haul flight will be peanuts compared to the data transmitted purely for entertainment. But even then, the satellite connection will only work as long as the plane can maintain some sort of stable attitude. For example if it was flipped upside down, the satellite transmitter would point in the completely opposite direction. The pivot mount and phased array antenna can only adjust/track for a certain number of degrees, not 360° in all axis. And probably not fast enough to keep up with extreme maneuvers either (especially roll). So keeping the black boxes will always make sense. But as more and more data is being streamed even just for the benefits provided to fleet management and technical maintenance, we will most likely see almost zero cases of completely "lost" aircraft, because there will be some data available up to the point where the aircraft enters such a violent attitude that impact/breakup is likely only seconds away. (Well, unless we get another Air Astana 1388...)
The amount of telemetry in a modern aircraft is astounding compared to the rather minuscule amount of data required for an audio stream. Modern codecs like Opus can encode voice at 56 kbps, which can be achieved even on an antiquated dial-up modem. And that's for real-time communication. If you don't care about a 100ms delay, you can do even better. A 1TB microSD card can store 4.5 years of 24/7 audio.
Like said in the video: Mostly you don't need that much data. Sending your location and some basic telemetry data via ADS-B and sending a burst of information when the system detects an anomaly sounds (like said in the video) sound more practical in my ears. I'm no expert at all, but transmitting all this mostly unneeded data seems like a huge waste of bandwidth. Also there is another aspect: I can imagine that pilots could feel really uncomfortable when all of their (private) talks in the cockpit would be transmitted and observed in real time. There is one rule in data collection: Collected data will be used and often it will be used for more then the intended purpose.
@@marc-andreservant201 Despite their name cockpit voice recorders record a lot more than just voices. If you use a codec designed for voice comms you will lose a lot of the ambient noise which can give clues what happened. The reason they are uncompressed is because you want to hear everything, at all frequency ranges, when studied at a forensic level. The moment you add any form of compression you can no longer trust what you're hearing is what actually happened.
40 or more years ago, the German Luftwaffe Tornado IDS aircraft was fitted with an ejectable (and if I remember correctly) combined ADR/CVR. The aerofoil-shaped device was fitted on the spine behind the cockpit and was ejected from the aircraft in the event of the wingtip or nose mounted crash-switches operating. The device was ejected into the airflow to clear the airframe and land away from the crashed aircraft. It was designed to float and contained a battery-powered transmitter to allow easy location after ejection. This technology has existed for years, so with the immense cost of trying to locate a sunken aircraft and understand why it was downed with implications for the remainder of the fleet, the system could easily be fitted to aircraft either during build or retrofitted. Yes, there's a cost, but economies of scale would apply and it just needs legislation to enable it. Modern technology would allow for more data storage with multiplexing and data-compression techniques plus rugged storage devices would survive fire, impact and water immersion. Basic ADS-B with positional data would keep track of the airframe if receiving stations could be provided to increase coverage, maybe with satellite or balloon systems. Adoption of such a system just needs the industry to realise that it costs a damn site more money to locate and find out why an aircraft has crashed or disappeared than it does to proactively pre-empt the problem. After all, aviation safety is no accident!
Great Job mentour and the team! The quality of your videos is just way above the bar, and it just keeps getting better day by day! I personally just want to say being a long time subscriber, i am loving these good old style technical deep dives like you used to do back in the day, And i do hope you continue doing them! And honestly just my opinion of course, but i think with how good the production quality on these videos are, you really could consider making long form documentaries on various aspects of aviation. You know like a proper truly deep dive documentary going into the history of it, and the backstory, with wonderful graphics and visuals. Similar to what you might see on some of these other aviation youtube channels like Mustard, and paper skies, just to name a few, who make these really high quality documentaries on specific aircrafts. I really think you could take this a step further, not just talking about aviation accidents, but full documentaries on various iconic aircrafts, and the development of technical systems like what you talked about here about the black boxes. Just my two cents, i think you could really become THE channel for all things Aviation with how well researched and detailed these videos are. And i must say you truly are a wonderful explainer! The detail and objective and thoughtful analysis you do is beyond anything i have seen anywhere on youtube! Anyway those are just a few wild ideas i had, Love all your content as always! And looking forward to seeing this channel grow to new heights!
It's quite simple. You just include a simple api to parse the uploads in stream segments just like exchanging ffmpeg or vlc inputs and outputs into groups of nodes. When one stutters, the other 8care there to determine the missing content, patch it, and match the input keys. Keys match = content successfully streamed into analyzable binaries. Binaries incredibly lightweight, non radio/transmission harming anymore than in flight wifi or carrier movie and show streaming on the client facing server. You could even take a playbook from Silicon Valley, and stream parts to the carrier app itself, and offload and uncompress the secure content on landing; parse it off to db server, remove from cache of app, and bam done. All within 2-5 seconds, if not under 10 milliseconds, and all under 150kb on the longest flight you can physically book (28 hours of black box data under 150kb) The tech is there. The investment in competent devs at the carriers in line with those at the FAA is the issue. The devs at Boeing and hardware/software backends are not the problem.
@@anon_148 How do you think carriers and streaming apps offload a 50-300GB movie or series into compressed chunks at 100-300% less size? They utilize timed compression layers. Slowly, processing enough data (buffer) a little bit past what the user needs, and nothing more, while slowly placing checkmarks on each chunk, so that no data is lost. (skipped) In this case, it could be on a pre-timed program throughout the flight. The entire Netflix library is hundreds of thousands of GB's. All at the top of your fingers, pre-playing away, utilizing various techniques to deploy that content to you at a moments notice. Placing the loads in to chunked client-server-serverless nodes allows that process to be lightweight and fast.
The ability of the pilot to transmit a portion of the data (e.g. latest 30 mins) could be useful for troubleshooting issues from the ground in certain situations
The idea about having a "burst backup" in case of any failure makes sense. Either way the system should have offline storage buffer in case the network is down, so a rolling backup of the offline buffer would be the way to go even if the entire blackbox contents were backed up when satellite internet gets cheaper. So that is a good first step. Also, lots of people underestimate how expensive and limited bandwidth satellite internet is. Transmitting gigabytes of data on the regular via satellite is still *very* pricy, for example on ships you usually buffer non-critical data during voyage to be transmitted via cheap networking in port. This will hopefully improve in the near future (with initiatives like Starlink).
@Alfred Weber he kind of did. It's a cost vs gain vs possible viable alternatives (manual/semiauto data bursts during abnormal situations) argument, which he did.
I just can’t agree. I wear a seatbelt despite the rare chance of an accident. But I can think of a lot of reasons why a pilot wouldn’t want idle chatter being uploaded to an easily accessible medium 🤷🏻🤨
As a person who monitored lots of sensors via modem, I say it's possible. As people who played online games most of you will agree. Because: 20G/2h ~= 166M/min which looks significant. And it's uncompressed (or only lightly compressed because it makes post-mortem decoding harder, but this is unconfirmed!). BUT you don't have to stream audio! It is already transmitted over analog radio! I'd say transmitting realtime critical sensor metrics is more than enough in emergency situations. To make a fair analogy: a multiplayer game (which is sensitive to time lag) transmits tons of parameters of any kind from boolean on/off flags to floating-point coordinates, and we played FPShooters on a dial-up! It may also help people on-gound to see the control panel and help the crew right away if possible.
Petter, you did mention in one of your recent videos that aviation rules are written in blood. Just when regulators think they've seen everything, something new and unexpected comes up that forces the hand of the regulators once again. We don't have the CVR and FDR of flight MH370 but there are strong pointers to it being a pilot suicide- one of the most difficult to counter problems in aviation simply because there are little or no warning signs. In the case of the flight MH370, the existence of a simulator flight plan on the computer of the Captain who flew MH370 which ended up in the Indian Ocean that matches the estimated flight path by means of doppler satellite pings, the recent revelation that the plane went into a holding pattern at the end of the flight etc show a pilot fully conscious of his actions- trying to redress some unaddressed grievance in the most extreme manner by plunging a plane full of passengers and himself deep into some of the most remote territory on earth. I'm all for the old technology but what about the future? Mass Gun shootings were once quite rare in the United States but these days they're happening with alarming regularity. If pilot suicides become more frequent, we're going to see more calls for real time uplink of data, no matter what the cost. If I were the regulator, I would mandate live streaming atleast for flights over open water because recovery from ocean depths is often quite time consuming and very expensive.
the problem with the pilot is he had over 30k hours i believe, he had a family, he knew aviation inside and out, more than you and me combined. i dont think a pilot like that would consider harming his passengers for political reasons. The holding pattern indicates that too, because if he was in a holding pattern deliberately then that means he was trying to do two things at once: either defend himself from a crew member or a passenger, or maybe it was a failure in the aircraft that they just couldn’t figure out.
You know that youtube live streamings have delays, or how about live news reporting are actually prescripted because of the 10 seconds delay. Streaming videos is already a bandwidth hog costly. And you're talking about flight data that we don't know how big it is. Aircraft manufacturers know that airliners know that it would become more costly to operate such feature and therefore even the tickets could not cover the cost of one operation.
Hi, love the content 🙂, just wanted to mention that the background music is quite a lot distracting. I know, many people do not notice it and dont mind, but it is one more reason not to use it at all. In some short sequence for dramatic effect background music is great, but through the whole 17 minute video it is too much. Thanks for the great content anyways!
I agree. But even worse, the sudden sound effects are disturbing as well without adding any value at all. Same for some flashy transitions and these strange glitches in between. I rather get the impression that someone is playing around with the cool effects of his video editing software.
With things like Starlink, I imagine it will be far easier in the near future for aircraft to stream information, though it should be clear that anything like this should be _in addition_ to a physical black box, not instead of. And as other commenters have said, having e.g. 1TB of solid state storage, in addition to what they already have, would be a cheap addition.
My day job is IT Petter and I agree with you completely for all the same reasons. An uplink is only going to be useful for those rare occasions where the FDR and CDR are destroyed.
Your informative video on the black boxes clears up many discussions on replacing these boxes. In the conclusion of this video the short blast theory, sending an immediate help signal, needed is the best idea. For all the reason given the short blast ability to send immediate call for assistance would keep those monitoring the continuous volume of data from being desensitized from information overload.😮
The cost of streaming black box data via satellite should be contrasted with a) the cost of searching lost black boxes from crashed aircraft and b) the cost of the data that is lost in case the black boxes aren't recovered. Especially in case b) the lost data is almost invaluable so it seems like the expense would be pretty easy to justify. It is rare for the black boxes to remain undiscovered but we don't know what caused MH370 so it could happen again any time.
If you consider how many flights there are every day, and how many crashes where we don't find the black box, I think that there is absolutely no way that the cost of finding the boxes comes even close to the infrastructure costs.
@@jsfyxzuf117 if, for instance, the entirety of Sikorsky's in-flight data were to be duplicated to Bell in real time the industrial espionage would be worth millions, then there's the possibility of diverting & corrupting data in transit, for example adding a tail-rotor vibration to every example of a rival's model in turn so that the entire marque is grounded till the fault is found. The scope for misdemeanor is endless, security must be very carefully considered & implemented in the initial concept.
20GB transfer over the satellite seems like a tiny cost compared to a cost of a single flight. Especially if it's bought b2b in massive volumes and internet connection is becoming a standard on planes anyways. But even that emergency burst can be virtually unlimited in length - start streaming current data + best effort stream older data backwards as far as it goes until the plane lands safely or transfer is disabled by a pilot (last part is questionable considering the china accident). In a long term if this becomes reliable enough it can even replace the black boxes as we know them and be a net saving.
Sounds like you've never paid for satellite time (or data.) I'm not talking about Starlink or HughesNet. Go price Iridium and BGAN. (still have a BGAN unit. that's a SIM worth it's weight in anti-matter.)
Yeah, good luck convincing US carriers to spend more on this technology. They rather find unique ways to cram more people into the cabin and operate at the bare minimum service levels than worry about the "just in case" scenarios.
Absolutely any system must only supplement the black boxes. Any system needs power yes but to be truly useful it must be isolated from the crew so that it cannot be turned off. As a minimum an intermittent location, altitude, course broadcast would be useful as a defence against rogue aircrew, or hijack or even just to find the thing. An early notice of a deviation from flight plan might also help in emergency actions (eg rescue).
The thing you are asking for is called ADS-B and is already mandatory on commercial flights Also black boxes now come with their own power supply able to provide electricity exclusively for them in the case of a loss of power (already extremely unlikely since the aircraft has one generator per engine, another one for the APU, the batteries and then the Air Driven Generator, a windmill that automatically extends when all of the other power sources are out) Source : I'm an aircraft maintenance engineer specialized in avionics
A “Black Box” is a common term in the electronics industry. Whenever you need to use something you are not allowed to or don’t need to know how it works, it’s considered as a “Black Box”. This could be a module designed and built by a third party which keeps the inner workings a company secret or a module which can have different functionality and or is based on different technology independently of your design. You have a set of specifications of incoming and outgoing signals you need to fulfill in your design. An example would be the CI (Common interface) slot on a TV. This was used for Pay TV before internet streaming. For the TV manufacturer, the card reader the customer can plug in is considered as a “Black Box”. The manufacturer of the TV doesn’t need to know which kind of Pay TV someone would use with the TV. The manufacturer of the Pay TV system on the other hand would consider the entire TV as a “Black Box”. When you are using a “Black Box” in your design, you need to make sure that all your signals towards the Black Box are within the specifications given by the manufacturer of the Black Box. You can’t check your design by observing the Black Box if it is functioning as expected to validate your signals. If the design of the Black Box is changed and your signals are not within the specs, the Black Box might malfunction or stop working. Test equipment and monitoring equipment is commonly considered as a “Black Box” by the developer of a circuit. It is natural that the avionics developer will consider the FDR as a Black Box. All the avionics developer needs is the specifications of the dimensions, power consumption, type of plugs, pinout of the plugs and the specs of the signals going into the FDR. How it works doesn’t matter. Knowing the true specs of what the FDR could handle can even be dangerous by being able to cut corners in the design. So it is natural that the airplane manufacturer refers to the FDR as a “Black Box”. Also there are two kinds of quality management. “Black Box Testing” and “White Box Testing”. “White Box Testing” is knowing how the component (Hardware and/or software) is made and how it works. Then you think up ways to fool it or make it fail within certain parameters. In “Black Box Testing”, you don’t know how it works so you can only randomly apply data or signals and observe its reaction. So my best guess is that crash investigators and/or the airplane manufacturer referred to the FDR as a “Black box” to the press because this is what they consider it to be. Something they don’t need to make the airplane work but they need to feed it specific signals and power in a certain way. How it works and stores the data and if it is certified or not is the problem of the FDR manufacturer.
As others have stated and talked about, there is no expectation that this would be a replacement for black boxes, and the figure of the data is in no means out of the realm of possibility. Being a software engineer (as one of the top comments is as well), I can agree that the amount of data here is actually near nothing in reality, and this isn't a system that needs to always be on. If we look at it as a system that is not a replacement, but an addition to the black boxes, we can omit or compress some data. Things such as air speed may be measured in 32 to 64 bit values, however 16 bit values here may be completely plenty for instance for its purpose, hell, we can send the difference in data with periodic full values to save bandwidth as well. This is also a system that can *conditionally* send data if any system has detected an issue, for instance if any warning is detected, out of normal vibrations, ect., the last X seconds, or minutes of data can additionally be sent for getting the conditions behind such an issue. Additionally, traffic is not as expensive as *storage* of data. Regardless of method (unless you're doing something crazy like sending data to and from the moon). This is a cost that can be mitigated by simply deleting data after as long as a week, or having the dataset reduced based on select parameters (master caution may retain data for longer for instance). All in all, it is a shame that this system has not yet been put into service, there are no excuses that we have not been able to put technology so readily available into planes, especially when most of the tech required for such a system is already in use for giving passengers in flight wifi....
In many (most?) cases the most important data retrieved from a black box is not in recording the catastrophic final moments where a satellite connection would likely be broken. It is in finding clues as to to the cause of the accident by analysing voice/data in the run up to it.
I have the feeling, that in aviation always something has to happen, until something happens! Can you talk about instances, where engineers thougt through systems, procedures and improved them before something happened? Your great videos on the Mentour Pilot channel sometimes give me that negative connotation, that always something has to happen until things get improved!
That "burst transmission" idea is what I was typing up while watching this video, because compression and burst transmission cut a lot of the overhead of "streaming" down. And no, replacement of black boxes sounds like a bad idea. Augmenting them with extra capabilities like burst transmission sounds great. If we can get it back to 5 minutes of data in that burst, that probably takes care of most of the technical incidents I've seen on this channel, because they seem to occur very quickly.
Arguing against redundancy seems a little silly to me. Especially since it's with something this relatively simple. Split the data stream, including voice recorder and send one copy off the vehicle in an asynchronous manner. More data=more better. It doesn't have to be perfect, and the cost is not going to be that high as long as you're ok with a best effort channel rather than a dedicated guaranteed data channel. 20GB over a 4 hour flight is "only" 12Mb/s. Way less than what the in flight entertainment systems use. Besides. Data can be compressed. I would be very surprised if you couldn't get the data stream down to under 1-2 Mb/s. Just my two cents
I agree. It should not be a question of one or the other. Do both. If data stream was cut, there is still the chance that the black box could be recovered. Starlink will definitely make it more likely to be doable.
Turning 12mb into 1-2mb isn't really possible unless the data is very repetitive. Plus, compression takes CPU power and time, which means that if it was compressed it would take longer to be ready to be sent off to servers.
@@YHDiamond And luckily, audio is very compressible. MP3, for example, has an 11:1 ratio. More application-specific methods could be used to compress things like the aircraft's position, speed, and other things. As for CPU power and time, by watching this YT video, you forced your computer to perform some pretty extreme compression and decompression, in real time. Even on an a VM on an old chromebook, with a Celeron CPU and other stuff in the background, I was able to compress stuff at bandwidths way higher than 12 Mbps, so it would easily be possible with modest computing power.
@@YHDiamond so i just did some back of napkin calculations. 88 channels, with a sample rate of 1000Hz, half with 8 bit precision and half with 16 bit precision is just a hair over 1Mb/s. Add 4 channels of voice data at 64K and we're still way under 2Mb/s. And yes. I know there are a lot of assumptions in my calculations. But i don't think that they're that far off. And I'm pretty sure that most data is repetitive enough to be compressed. You don't really need 1000 discrete samples of engine speed. I'm fairly certain that a compressed second would be more than adequate. But again. That is just my assumption and i realize that
My guess is with thousands of planes in the sky at any given moment, the amount of data to deal with would just be phenomenal..... (commented before watching the video) PS I JUST booked my flight YUL to YVR for July 1st with my cat WOOHOO and i bought a vintage Air Canada tail fin pin off Ebay that I'm gonna wear then ❤️
@@c4fusion1 it's because the aircraft is already fitted with different satcom antenna. I guess you already know which one is it. And as a guy from a broadcast background, most videos you watch right know are encoded and serve you as compressed-ready-to-watch format (that's why the bandwidth cost are so small, and there's no interference in terms of watching experience), while flight data are mostly raw data from the sensors that even the most powerful smartphone couldn't decode it in seconds. And most of the time, on poor connections, video streaming would just buffer before it receives valid data. Flight data streaming does not work that way. If you experience packet loss/high ping due to insufficient wireless transfer rate then the received flight data becomes unreliable. Also adding tools like this means the operating cost of the aircraft increases.
Why does it have to be a different antenna if we are just using this as backup to ensure that we have data? We can use the internet to transfer the data, and before you say it's too sensitive, why would it be? The world's financial data is run through the internet. As for the bandwidth, how is it not enough? There is only 5 voice streams and 38 parameters. Let's say the 38 parameters need to be recorded at millisecond precision, it's still only 1.216 mbs (assuming the parameter fits in 4 byte, which if it's a number it will) or about tenth of the bandwidth for 1080p youtube video. As for compressing voice data in real time, a cellphone from the 2000s can do it digitally much less a modern cellphone (incidentally, a modern cellphone would probably have more processing power than the flight computers in most airplanes). As for the video streaming comparison, I wasn't making a point of what equipment you need (though you can live stream using h.254 via a smartphone these days) as much as how much bandwidth you need. Since satellite internet unlike cable internet is symmetrical, the down speed would be equal to the up speed. The nature of the internet As for the packet loss issue, there are many ways to ensure high availability of your connection such as packet retrying that is built into most streaming applications such as the youtube video you are watch. Even better yet, I know of a sewage treatment plant that has a satellite plant and pumping station that is controlled and recorded remotely. This being a backup would not need as high of a reliability and if it did, the tech is already available and is at least 10 years old.
Along with voice recorders, I feel like cockpit, cabin and external video recording should be a worthwhile next step to include in planes. They'd help investigate things like bird strikes, engine fires, pilot behaviour during an emergency, or even passengers being dangerous. It wouldn't just be useful for visually investigating a lot of issues before they escalate into crashes.
We do not need to stream all of the Black Box data to the cloud. We need only to send a small packet of the GPS location, speed, and direction every 30 seconds for aircraft over oceans or remote areas. If there is a crash that will help us find the Black Box more quickly and maybe survivors.
Teah. I support backups of data. I would have reccomended a smaller light version of a black box that maybe records the last 30 min of flight data and 10 min of audio, rather than the standard black box that records in 20 hours and 2 hours respectively. This lighter black box should be stored in a separate place of the plane with less protection from being damaged because, let's face it; making theese things really robust to the point where they can withstand direct impact with the ground is really expenside. A plane normally isn't 100% pulverized in a crash. There will be some pieces of debris that end up with minor to no damage. This light version is just there in case the regular black box had bad fortune and got destroyed, that it might hopefully could have the good fortune of being in a place where it does survive. There's still the posibility that the plane might land in the ocean and nothing be recovered. So a transfer of data through radio or simething is a good idea. But such data transfer is costly in large amounts. So, yes. Only small amounts of data bursts in case of the plane exceeding certain pre-determined parameters would be more than enough help for investigators to try and discover the cause of the problems. Anything is better than nothing.
I think it kind of goes without saying that a streaming flight data system would be supplemental, and not replace the black boxes. With systems like Starlink, the cost of net data transfer is going to continue to drop until it gets trivial. Especially in the most extreme examples (such as the recent nosedive of China Airlines 5735), a satellite link is probably going to lose contact due to the attitude of the aircraft.
as satellite internet decreases in price, the "cloud black box" will become the industry standard. Just a matter of time and economics. However, during the MH370 investigation, I remember there being a segment where they mentioned the engine manufacturers such as GE and in MH370 case Rolls Royce receive periodic updates as to engine performance from their control centre in the UK via Data link. infrastructure is in place, just need a "Henry Ford" to make it affordable for all.
That data link is quite different from anything that can support full flight and voice data livestreaming. Remember that for this to work you need 100% coverage, so having 'enough' satellites for full coverage is not enough, you'd need full overlap in case a satellite doesn't work or is overloaded etc. Even Starlink would not be capable of this currently, perhaps in another 10 years at the soonest. The amount of data, lets take the 20GB per flight Petter mentions as an average means that if we only keep a rolling 24 hour period there's plenty datacenters that can store that without issues, but getting the data from the aircraft to that datacenter, mid-flight is not possible right now, not 100% of the time anyway. Then the biggest issue, you mention 'just a matter of time and economics' but which economics? Where is this ever going to save money let alone make money? It's a hole in the ground which will never earn money. It perhaps could save part of the cost of searching for a lost aircraft as we would just know where it is but the entire cost of the system would be hundreds times more than those potential savings... This only saves money if we have a missing crashed aircraft every day or so.
@@someguy4915 once upon a time, we had dial up modems, having a cloud account like google drive or Dropbox seemed like an unnecessary construct. Downloading or uploading 100mb seemed like an eternity. At that time - if I told you that homes and apartment blocks would be connected with fibre optic cables, you would rightly call me insane! Yet, here we are.
@@someguy4915 with regard to the economics - off the top of my head, firms like Lockheed Martin or Raytheon or even Boeing need to start mining data for their autonomous flight programs. Whilst drones are just in the military spheres right now, in time it will filter down to commercial airliners as well.
@@ahmedmotala7282 Don't expect pilot-less airliners to become a thing within the next 30 or so years, public opinion would have to be changed first, then legislation and then they also have to build it. Besides that, I doubt that some blackbox data will really help in that field without the full image, the blackbox reporting a rough landing is useless without weather information at the time and without a report of the pilots explaining that there was a microburst that the weather-radar can't see etc. But even if this could be useful for data gathering, that's no reason to deploy it on all airlines, just on a select few where they can get the side-information from pilots etc. And if any single one of those aircraft crashes the first question the public will ask is if that new 'drone passenger plane' program had caused it, destroying the chances of such aircraft ever being accepted by passengers.
TL;DW: because, especially since crashes rarely happen, it's too much data (more than about 20gb worth per airplane per flight) that needs to be sent all the time across all flights. The last 20 hours of flight data and the last 2 hours of high quality voice communication need to be recorded in order to determine what happened in the event of a crash.
I’m a software engineer, every single request made to the web server is almost always logged, so one person opening one webpage could have 20+ long lines of logs, and each interaction with the web server is logged in one form or the other, these data are preserved for much longer than 20 hours.
I really like the short burst idea. With the position data combined with attitude, I'd imagine the crash site could be nailed down very accurately. What I don't get tho is why the cvr still only stores 2 hours. I get that it's probably not a grocery store level sd card in there, but still. Is this special storage that expensive?
Id imagine its probably something to do with making it durable. Can only fit so much tape with all the armor around it. This is just my best guess btw.
No it isn't. Even high quality flash used for missiles is dirt cheap compared to even the metal box that encloses the electronics. We actually fit much bigger flash than we need because the smaller ones aren't made much now. The night vision goggles I worked on had 8GB of flash. We could have stored a million seconds of phone quality uncompressed audio, or 200,000 seconds of high quality stereo. That's over 2,500 and over 500 hours respectively.
When film or tape was used it was a technical problem. When it got digital in solid state chips, it was no longer a technical problem, but a political problem. The pilots in USA has insisted for years that it should be no longer than 2 hours.
There was an accident with a cargo ship several years ago which steamed into a hurricane, by law they need a couple hours of data and voice recording on the bridge, but by sheer luck, a much larger SD card was used, so something like 15 hours of voice recording was available, and was very helpful to understand the decisions which led to the sinking. Of course, ships take much longer to get anywhere than planes, but records of an entire flight could be very helpful indeed.
@@a4d9 It was wire before that, because wire survives crashes and fires. All FDRs should be switched to solid state now, tape wears out too fast, and breaks easily. Wire was tough but couldn't record much.
My thoughts: tail is wagging the dog insofar as data recording goes. The solid, on-board works and takes decent amount of engineering to survive a physical shock, But, it is relatively obsolete. It should continue but should be buttressed by some modern technology. Even if the technology isn't 100%, This kind of data is retrospect and has nothing to do with flight control...it is forensic only, The more forensic data you have the more likely you'll arrive at a proper conclusion. Even if the 'real' black boxes drown, you'll have some data. Oh yea, FAA or whoever would say it needs to be 100% reliable and without gaps. But does it really need to be?
Given that the reputations of the living and the dead will be on the line they'd be right to insist that it be reliable and without gaps. We can aspire to Just Culture all we want, but unfortunately people in pain demand to know who to blame (even if they end up blaming the wrong person).
To me there seems to be a privacy reason, too. Imagine all flights are readily available this might encourage some airline managers (think Ryanair) might want to improve their performance by using that FDR data in an automated fashion from each trip. This could make it even more uncomfortable for pilots and generate useless stress as they could link everything to the actions of everyone.
That data is already available to the public. The only way one has privacy while exercising the privileges of a pilot certificate is maintaining visual flight rules, or in some areas flying an aircraft without an engine-driven electrical system under visual flight rules. How do you think you can follow aircraft by registration number on various websites?
@@r2db That is ADS-B data but as far as I understand it does not have as many parameters as raw FDR data. And of course you can see some performance data like F/L, speed and direction but (correct me if I am wrong) no real direct steering inputs and other data that are not sent by ADS-B. Like rudder/flap/thrust settings It is also only updated every half second which sounds like it does strip out lots of data. I think it's ok if anyone can follow a plane but if someone would look at every keyboard press I made in my job I would feel quite a bit more intimidated.
@@gentuxable My point is that your privacy concern is already a lost cause and has been for some time. It did not exist for aircraft on IFR flight plans pre-ADS-B and it doesn't exist now for any ADS-B capable aircraft. Some engines in many different industries have telemetry data that already communicates regularly if not continuously with the manufacturer (as an example, the Cummins X15 where it has the capabilities for over-the-air parameter updates). I have never flown an Airbus but it seems that pilots of their aircraft have been aware of telemetry data it sends for some time, and farm tractors such as some John Deere models also have OTA capabilities that allows the company to disable them if they are stolen or involved in some present wartime action. In healthcare, every mouse click and every keystroke is logged with a time stamp. As a pilot, if data is sent over the aircraft bus to the FDR then there is nothing preventing a separate piece of equipment from logging or transmitting such data for corporate purposes. If you do not want someone looking over every keypress you make then in today's environment the best suggestion would be not to take a job involving anything with a computer in any form.
@@r2db Ok I get what you say but to me it seems like it still goes within a reasonable range with ADS-B and clearly there are some companies that go far beyond what we outsiders know of. We're getting into dangerous waters in almost every aspect (which I'm sure would also lead to some ruling if court at least in Europe knew and that is often an issue already to bring it out). Having the FAA (or EASA for that matter) mandate all 88 FDR uploaded to a server seems to go further than everything done nowadays I'd still think and it would send undesirable signals to the industry. BTW if my company records every keyboard stroke I'll sue them, I know my rights and that is a clear violation.
Well it's already mentioned in mentour pilot channel regarding pilot privacy. And most pilots agree if their airlines use flight monitoring system like ACARS. Still wishing for privacy? You know that lavatories can be opened from the outside right?
The QARs you mentioned can do this. They send recorder data to an airline HQ via 4G cell service. If a pilot lands hard they know, an HQ can also use the data to save fuel (make better flight plans with different speeds). Also recorders still work during an electrical failure. The battery supplies essential power. However it may not broadcast anywhere since that works off main power.
Not really impressed by the data transmit volume argument. It could be cut into tenths and give a better clue than nothing at all.. Not crying for the airline's expense. They sell the same game to customers(internet). Really now. Data recorders are old and very good but, they're old. So am I. The data X-fer is not much given the tech now.
Well whether the cost is high or not the cost will be passed down to the customer and as it's so rare to need the black box and not have it, I'd rather not pay for the cost of sending the data I'm not going to need rather than paying for it on every flight just to have a less than 1% chance that's going to be the flight. I'm in the US and flights are already expensive enough here. I think a better case could be for private aircraft since they have way less data to send, but the data we could get out of that could be just as valuable.
@@test40323 No reason it can't be, other than nobody having really asked for it yet. All this about ADSB is just a sideshow, there are plenty of communications satellites, they just need to work out what they want. But anything requiring pilot intervention is a non-starter. There was a great interview with Chuck Yaeger about being a test pilot and the bit I remember was him describing seeing a guy in a crashing plane "The guy is running all over the cockpit writing notes and talking on the radio. That's all BS. You are just trying to save your life." That's the way it is.
@@cageordie , good point. But some emergencies have more time and if the ground crews can help figure out how to resolve it based on the live data ... the work load would be eased by the extended team.
In two words: Data Analytics A plausible reason why air lines would adopt full flight data upload to the cloud is regulation. If it were regulated this way then the airlines would not have a choice. In this case the cost of the data uploads would be reduced by economies of scale. If I had my way, safety orientated systems would be charged at cost. The question why would you do this and you can't just say Data Analytics? The main reason I would suggest is that with each flight storing it's data in the cloud, historic flight data could be studied. In that way, in the event of the loss of an air craft whether the Black Box is recovered or not, the flight data is able to be studied to try to determine where a fault first showed signs of developing before the incident. Or alternatively, when pilots begin reporting a new fault, then the current data may be studied and correlated between aircraft to detect the fault and trigger pre-emptive actions on air craft where the fault hasn't yet been reported. Other studies of historic flight data are available. P.s. This or something similar may have been mentioned in an earlier comment but I haven't read all 1000+ comments already posted.
I am an electronics engineer and a pilot, and i completely agree with your analysis! We will need a new satellite network just to track all data from all planes out in the sky all over the world at the same time...
Yeah, especially voice data needs really much bandwidth. How much data (and which data?) is transmitted via ADS-B? I guess with increasing bandwidth availability more parameters could be transmitted this way.
@@annando ADSB data is transmitted only when the transponder is interrogated by ground radar or other transponder. Usually between 1 per few seconds up to few times per second. It requires low bandwidth, since transmitting very few parameters , at very low rate.
@Oferb553 BTW: I just saw the latest video by Scott Manley and coincidentally he spoke about ADS-B. He mentioned that Canada now requires a space based ADS-B for planes in their air space.
As a complete casual and non-expert (even "amateur" might be pushing it...), what I find disconcerting about aviation isn't the black box issues. As it was said, they are quite reliable in most cases. But what is apparent to me in a lot of the videos here and elsewhere, is how unreliable the day-to-day voice communications are. From relying on English globally (or any one language no matter what it is), to either controllers or pilots missing an instruction or getting it wrong, or having interference garble a transmission, it feels like there's got to be a better way to transmit and acknowledge instructions more reliably.
Very true! At the very least, there should be voice recognition software in every cockpit so that instructions are not only heard through the headphones, but also appear as text on a screen. Ask any pilot whose first language isn't English, and who has struggled to comprehend instructions rattled off by a guy in a control tower at 1,000 words per minute (or thereabouts.)
@@nicolasuribestanko A text message is a good idea, but voice recognition is not. Voice recognition are not reliable at all and could cause a lot of confusion if anyone was relying on it and it transcribed wrong, which it definitely will, which will cause serious issues for safety when it comes to receiving instruction from ATC.
What is far worse than using one standard language, with a standard set of phrases? Not having standards. I earned my pilot certificate in an era of "position and hold" and still dislike "line up and wait" but it is what it is and I can't change it. If I read back the instruction as "position and hold" I am not properly communicating. Why is voice communication still the gold standard? Because when one is flying an aircraft, particularly during an emergency, one does not have time to type on a keyboard. Both hands and both feet are frequently in use during an emergency, and voice communication takes one finger with an occasional twist of a knob or button push to change frequencies. There is also no space in the panel of an average general aviation aircraft, and definitely not in the flight deck of a heavy transport aircraft, for a bunch of additional buttons for communication.
Cloud storage could be a backup, not a replacement. The backup stream could focus on just the changes or the most critical parts and with the latest compression algorithm this is definitely possible to do. They should have a cam in the cockpit and a tracking device on the black box. Everything can always be improved.
We don't need to have real-time transmission of data and voice record to replace the black boxes all the time. I agree that operational and cost consideration must be balanced with safety needs. At the end of the day, we cannot and do not aim at eliminating ALL risks. We minimize it with minimal cost and operational difficulties. The "operational data monitor program" you talked about can be further enhanced. We can also make it to be triggered by alert events (e.g. when a master caution or GPWS warning or stall warning is issued, etc., or even when the voice recording detects the pilots said "PAN PAN" or "MAYDAY" in the transmission to the ground, etc.). We can also make that "short burst of data transmission" to contain more data (e.g. contain more parameters and even voice recording, or with data in the last 30 min instead of 2 min). Lastly, we can also make the transmission real-time since the alert happens, so the "cloud" will receive continuous data since the alert happens (instead of just a short burst of history for a few min before the alert) for the next 30 min. Things can be improved this way further, rather than a complete replacement. We actually rarely replace all technologies completely. For example, we still have telegram and fax available, and bank branches are still everywhere despite online and mobile banking, etc.
I'm not sure, to answer the question. If I had to guess, it would be $.... In any case, over the Indian ocean there is no radar observation. But....Over the Atlantic? Despite the fact that all the airplanes use CPDLC to communicate their position in space to Gander/Shanwick, there ~~IS~~ radar coverage over the north Atlantic! This is specialized air defense radar, and we don't want the filth in Russia to know it's capabilities, so your plane is officially not in radar contact. Even if it actually is...
The size of data needed to be sent like you said 20 gigabytes for example could be reduced massively by using some new file compression tech. Probably down to single gigabytes and then when it would be viewed it could be decompressed to have the original quality and size. I think the reason streaming black box data hasn't been implemented is just because companies are greedy to make money and want to avoid improving anything that isn't absolutely necessary by the law.
20GB is not a lot for LEO satellite constellations like starlink. That’s probably on the order of a couple dollars. Plus each flight can have hundreds of tickets to pay for that
The housings for commercial & military black boxes used to be made by Valley Aluminum Works in Phoenix, Arizona. They were made from hot forged aerospace grade aluminum. From VAW they went to Harvey's Deburring, then to a machine shop, back to Harvey's for more deburring, & degreasing and then on to an enamel powder coat paint shop for a hot baked on orange enamel coating.
Get 20% discount on the yearly subscription of Brilliant by using this code 👉🏻 brilliant.org/MentourNow/
Elon Musk's Starlink satelite internet is ideal for streaming both real time video, CVR and all data from the aircraft, throughout the flight!!!
Not implying 'full replacement' here..
but the 'high cost of transmission' and 'possible zero use of the data in most cases' (since no incident)...
Does *NOT* make for a strong case.
There are literal behemoth (to name a few; ABC Inc./ Google, Amazon, Meta/Facebook and zillion others) whose businesses routinely have massive intercontinental data transfers.
Of-course a physical unit, like you mentioned.. and is commonly known... presently is the only way of ensuring that most data till the *end😪* gets captured in almost all cases.
But whenever watching documentaries or coverage.. where they ever said... oh but the CVRs were only designed to have preceding N minutes of data...
I often felt a mix of helplessness and being infuriated.
Data costs: nope, disagree
Cost on environment or impact of server farms, yes perhaps.
But (to me) it still feels.. that in our lifetimes.. whether due to regulation or need based 'choice'... fixed wing aircraft... and certainly so helicopters and such.. operating in difficult terrain.. would have it as a 'standard feature'.
It = not just 30 seconds - 120 second bursts of data relay...
...but ongoing continual transmission of key parameters.
So automated and autonomous to the extent that even some on a suicide mission won't be able to disarmament or disable.
Triple, quadruple redundancies... including separate ways to power the electronics or avionics which perhaps sustain for a while even when engines are not working.
@@sailaab Your analogy of the "behemoth" companies that have huge intercontinental data transfers doesn't really apply to airplanes. Those companies typically use either landlines or shipping companies ("Never underestimate the bandwidth of a station wagon full of magnetic tapes hurtling down the highway" -- old joke from the 1970's that's still relevant today) to move data, neither of which are typically available to airborne aircraft.
Even when the big companies do use wireless means (satellites) to transfer data, it's one transmitter to one satellite to one receiver. There's a lot of overhead in having a lot of different stations on one network; just ask anyone who's tried to use the Wifi at a busy coffee shop.
I'm not saying it's impossible, just _really_ expensive; and since over 99% of black boxes are recovered intact and usable, there's just not a big demand for it.
@@Seventh7Art ججججج
IF they add a system like Starlink, They can not only send airplane data but 4k video from the copit to their private servers. So they can be used in a case of problems.... CCTV in a plane... but it will be horible to watch the final minutes of the plane and people...
I'm a software engineer by trade. The figure of 20GB stated in the video is probably a reference to *uncompressed* data, and with modern data compression algorithms, can easily be reduced to 2 to 5GB in practice, and can be handled by an LEO Satellite constellation. It may not be necessary for every flight type where accident debris would be left on the ground, but could be accomodated for transoceanic flights, like MH370. Think of money spent for months of fruitless ocean searches to find that plane, and the cost isn't nearly as onerous. The number of parameters sent such as simple latitude, longitude, altitude could make recovery of the physical black boxes a possibility as well, and should also be less expensive. In short, there is no reason other than corporate inertia to not attempt to stream "black box" data to satellite, then the cloud.
Yeah, if you can literally store data on a disk you are largely unconstrained about what you can record and store. But if you look at something like video game networking, they really just transfer inputs across the wire with each game client replaying a simulation which is not even a kb/s of data. You can downsamble, compress, and prioritize to bring this down to virtually nothing. There's also only about 100,000 commercial flights per day in the world which is an absolutely trivial amount of data to handle.
Absolutely. We're not even talking here about needing a perfect record of all inputs and voice data. There are many ways to truncate the data so streaming/storage is manageable *before* we have to consider data compression. As you say prioritising data for high-risk over-water flights might be a good start, and then we might choose to prioritise information that allowed us to find a missing black box. Even constant telemetry data could be trimmed to prioritised controls. Once a flight lands, there isn't even any need for storage. This is far from an intractable problem.
My feeling is that the real reason is that the *initial* cost of developing a robust system would be significant and no company really wants to pony up the initial outlay.
Also a software engineer by trade - compression works great when you can leverage the patterns in the adjacent data to de-duplicate stuff and save space, but its a LOT harder when you are talking about real-time data. I don't even mean real-time like a movie video decoder, but real-time like aviation and battlefield systems process that are so tight a loop you can't even use a normal OS but instead use customized stuff like RedHawk and VXWorks that can ensure there's zero delay in processing time. Its also possible the 20GB figure might already be compressed (he doesn't say in the video, and we don't know for sure). I've not found any examples in my job where real-time data processing is compressed on-the-wire, even if its compressed at-rest.
They can just send part of the data. They can send something and not just say "expensive"
@@TheDiner50 Well my thought is that sending data burst once-per-second would make it far more achievable than trying to fire off 20gb all at once.
I think the real point is, NOT a full replacement but a parallel recording system
exactly. the idea of keeping your only backup of data on-premises would make any IT professional cringe. the amounts of data mentioned are frankly trivial in terms of modern bandwidth and storage capabilities.
I worked at Satelliite and Wireless company for many years, and in 2001 was on a project to provide data links to aircraft. After the Sept 11 attack, we proposed that the data link also carry camera information. We did a video conference call with a Congressional Committee from the test aircraft to show viability. They said "great idea, now who's gonna pay for it"....and there it ended ....
Just shows that it's possible plus there's so the powers that be wanna hide from the public!
Wtf @@GG_Booboo
"Most black boxes are able to be recovered so why bother backing up the contents?"
Most airplanes don't crash, and yet when they do we take steps to avoid them in the future anyway. The entire ethos of the aviation industry is that the 'same' crash should never happen more than once. The fact that there's been more than one plane whose black boxes couldn't be recovered should be viewed as a failure of the industry.
"It's expensive :(((("
I'm sure all the million other safety features on a plane are expensive too, but they're all still there anyway. That's why we have aviation safety regulators like the FAA - to prevent people from just cheaping out on safety equipment.
Just think how many crashes may even be able to be prevented if the black boxes could transmit in real time during an emergency to the ground and they can possibly tell the crew where the problem is and maybe solve the problem enough to be able to land the plane. That could save lots of lives.
@@KevinBenecke while rare, you know also need to worry about tampering
@@KevinBenecke I think it takes the ground to long to answer, if it is important. Else the plane itself would warn you/send an error code.
If you have that time, then you could also have easier way for the most parts. ("Land" on a airport, with the firefighters ready and so on)
Imagine if the top 20 richest would die in plane crashes. I'm sure $20 billion to protect them all the way to the ground would still be "pocket change" anyway then.
@@KevinBenecke While reasonable on paper, I think something like that could be counterproductive. From this channel, I've gathered that in the event of an emergency, air control keeps OFF the radio as much as possible with the plane in distress because they know the pilots are extremely stressed and focusing all their efforts on getting the plane back under control. Interruptions are kept to a minimum.
I think that they should keep recording data locally on the aircraft as a backup when or if they start streaming the data. That way (if the device's data is recoverable), it can fill in data gaps from loss of live communication post-incident.
The thing is if they can start streaming the data when an emergency happens, they may even be able to prevent the crash from even happening. Ground control could possibly discover the problem while in flight and direct the crew on how to resolve the problem enough to be able to get to an airport and land the plain. The transmitting doesn't have to begin until there is an emergency. There should be a button in the cockpit the pilot can press to start the streaming. Plus there should be a system that automatically starts the streaming if a problem is detected which could help to prevent the problem from even happening because ground control could radio to plane and let them know there is a problem happening to which they can get to the problem or land the plane before the problem gets worse. That could save lots of lives instead of waiting until after the crash happened. A lot of the crashes that have happened most likely could have been prevented if they could do this.
the pilots already have alarms and an entire dash of performance gauges in the cockpit for them to know a problem is happening... what you're really asking for here is a check-engine light so ground engineers can watch your 80+ parameters remotely on a screen, never look at or inspect anything, and reliably find a solution based entirely on that - in real-time... don't think that's going to work out
@KevinBenecke True, an ai could be monitoring the data stream and bring up an alert to a human (eg ATC) when it notices something starting to go wrong.
Agreed. No reason in today's day and age not to just have both for redundancy. In fact I'm surprised it isn't mandatory for intercontinental flights.
Gaps in the data? Find the box. If no gaps? Well no need to spend god knows how much to go searching in the ocean for an incredibly tiny piece of wreckage that could be anywhere in a huge radius.
Seriously it can't be that hard. Ya know those planes are streaming more internet.
@@stackflow343more brains on the problem means more chances of solving it
While it is fairly infrequent that the black box is missing or too damaged, in an awful lot of your accident report videos the pilots forgot to pull the voice recorder breaker so we don't have what was actually said for the investigation. This could also be easily fixed by having about 1tb of solid state flash in the black box, to allow for months of recording to be stored, and would also be fixed by having a 700kB connection to a satellite (that a lot of planes already have anyway) and storing it off site
this is what i dont get. if there is any reason why the blackbox cant store so much data, just save the data secondary normally on any hdd/ssd inside the plane. The data is lost in an emergency situation, but atleast in most other situations the data can be saved and recovered
@@Perfektionist most SSD and HDDs aren’t that reliable in a crash.
@@Knirin i wasnt clear. I ment, that there should be a secondary equipment that is not crash safe. But this one could store more data
'forgot'
@@Perfektionist We already have that in the QAR.
For the record "the cloud" IS a server on the ground...lol.
But where you are directly transmitting to, from the perspective of the aircraft, may often be a satellite instead of a ground antenna, where it goes from there is less of an issue.
@@subbss it wouldn't be an issue unless the aircraft diving towards ground. Then it can't transmit any data to the satellites anymore. For further arguments, there's already been discussion in the comment sections for the why's.
Sometimes clouds touch the ground
Imagine flying an aircraft through the clouds and crashing with all the servers located there.
@@Aranimda 🤣
To me one major benefit to streaming this data (in addition to the black boxes) is that it could enable more long term storage of this data. In a lot of your crash Investigation videos there was critical information missing because it was more than 2 hours old. With long term storage you could look at older information even from different flights which might help draw better conclusions.
Additionally, there is a huge potential for algorithmic data analysis to predict failures before they occur. I'm assuming the airplanes have their own diagnostic systems already, but in industrial settings long term data is incredibly valuable as you can pick up on very minor changes and use it to prevent costly machine stops.
I mean it would be way cheaper just to double/triple the storage in the box.
@@philipwhiuk it really isn't. Data storage is complicated when it's still supposed to work after being pulverized along with the plane... In fact the very first example mentioned showcases this.
You are totally right: it will be possible to have supercomputers running an AI to detect potential failures on the ground. It is just a matter of time: sooner or later the costs will drop down and for the companies, the savings in maintenance and reparations will overcome the cost of transmitting data..
Flight data is analysed for maintenance purposes and is important part of the reliability. The amount of data recorded is only going up as we see the benefit.
@@philipwhiuk No, not nowadays. Cloud solutions nowadays pretty much outprice physical storage across the board
It was not the outside of the boxes that had to be painted black when they were using photographic film, it was the inside of the box that was painted black, to absorb any stray light. Black box was a term that originated in the theatre/film industry to refer to a stage or set that was painted totally black to absorb any stray light. The term was later applied to any container painted black inside to absorb light and protect film. As the early flight data recorders were using photographic film, they were painted black inside and hence were referred to as black boxes.
If a box would have to be painted black on the outside then i would highly question its Usability and the sanity of the person creating it.
Most of the settings will be relatively easy to store - you only store deltas. Whenever a switch is changed, you get a 3-byte (or thereabouts) change. Most switches are toggled less than 10 times per flight, so all of those fit within 3 kilobyte - over the full 20 hour timeframe. Those that measure a constant signal will record whenever they update - which is typically a max of 60x per second, with an accuracy of at most 32 bits. They are also typically ideal for Huffman coding - flight height does not change immensely quickly very often, and nearly all deltas will be very close to zero.
Then the audio itself. If you assume the audio channels will be compressed to 384kbit/s (assuming 6 channels at DVD equivalent quality) you should send a packet of ~50kbyte per second to keep up, of which 96%+ is audio. Using variable bitrates should bring that down to ~20kbyte per second given that most of the time, most of the channels are effectively silent. That becomes 72MB per hour - very very doable with satellite, but still too expensive to actually do.
Only upload data when the avionics detect an anomaly. 30 minutes of data = 36MB on a small percentage of flights = quite affordable especially with how cheap it's getting now.
Also a switch change only has to be 1/8 of a byte unless you're monitoring the analog voltage to see deeper into the wiring, slight arcing, etc.
@@thorlancaster5641 the con is, how would you know what were the pilots reactions and all stuff in the case of a sudden depressurization caused for example, by a bomb ?
@@User-jr7vf you wouldn’t, but you wouldn’t know that with a modern black box system if the bomb destroyed it either.
And unlike a black box you wouldn’t guarantee it’s loss if a plane crashed in the ocean. Which is a much more likely scenario then a bombing.
Internet is not new on flights, and Elon musk’s space internet is also cheap..
I worked on the Navy P3 Orion aircraft in the 1970's. It had a flight recorder mounted in the tail which will be ejected in the event of a crash. How do I know this? I accidently ejected one while I was doing some work in the cockpit when I bumped the switch. It successfully ejected and landed on the ramp. Of course my supervisor gave me the task of reinstalling it in the tail. It was an embarrassing event but memorable. The P3 also has "crash switches" in the nose that will trigger the eject mechanism automatically during a crash. The recorder is mounted in a fiberglass airfoil that can "fly" to the ground and it also floats. So this technology has already been around for awhile.
Awesome!
Gary, typically, black boxes are crazy heavy. Are the ejectable type lightweight? If so, can they survive staying on board a crashed aircraft if the ejection system fails? If aircraft power has failed (except, say, the battery), can it still eject?
On many civil aircraft, the recorders even stop working when main bus power has been lost.
@@ImperrfectStranger This recorder was not super heavy and was mounted inside the fiberglass airfoil. I don't know if it would survive without ejection. I suppose it depends on where the tail ended up at and if it was exposed to a post-crash fire. But with the crash detection switches being in the nose, that's usually the first point of impact so chances are good it would be ejected before a power failure. Unless of course there was a power failure before the crash.
@@ImperrfectStranger Hence the pitiful state of accident investigation data. Many times in the past accident investigation was delayed and hampered by the fact that the recorders simply failed due to lack of maintenance. This has been improved by digital recording, removing the delicate moving parts from the system, but really, how hard would it be to add extra capacitors to allow recording to continue, even for a few seconds after power is lost? Another item that should be added is a non-silenceable alarm that is triggered by removing power from the recorders, and instantly turning on the emergency locator beacon(s) of the aircraft if this happens. That would prevent "accidental" failures of the black box circuit breakers.
I will leave it to the reader to investigate why none of this has been accomplished on passenger and freight aircraft.
Doesn't ejecting the Black box before the impact of the plane makes us lose some data? Not even sure if the Flight Data of the "tumbling" is useful, but it sure does look interesting in some documentations.
I like the words "uploading the data-box to the cloud" to refer to a phyisical box being ejected
Aireon is already doing this. In fact, the FAA gets real-time data from the 737 max fleet via the Aireon network to monitor for issues, although the data points aren't nearly as comprehensive as the black box.
Starlink does not capture adsb. However Aireon is deployed on the fleet of next-generation iridium satellites.
Starlink does not capture adsb right now yes, however the aircraft could have software installed to send the data out over it's Starlink link for Starlink equipped aircraft.
Folks, SDR doesn't change the antennas or amplifiers. Starlink has Ka, Ku, and E band support. Adsb uses 1090mhz. Won't work without the right antenna and RF front end.
SDR means protocols can change but doesn't mean software defined physics.
Aerospace software/data/cloud engineer here.
ADS-B is all very well unless the cockpit crew disable it (as appears to have happened with MH370) due to a rogue pilot, pilot error or hijacker duress. Also, the cited data volume assumes no compression. Cockpit data are VERY compressible, potentially down to about 1% of raw data volume, which makes the costs of routine transmission and storage fall into the noise.
Those data are an absolute gold mine. By comparing similar flights on the same airframe over time, it may be possible to identify and predict failure. By analysing over a whole fleet, even over all airframes of a given type, there are opportunities to improve everything from design to training. You can even look at accelerometer data over time, correlated with GPS, to get fine-grained analysis of meteorological trends.
There is a saying in cloud technologies and data analysis: there is no such thing as too many data, only the wrong type. The potential for streamed aircraft data goes so far beyond crash investigation that that becomes almost a sideshow.
It would be handy to have this data stored on a server. In the event of a crash, investigators would be able to immediately access vital information while the other investigators go to the site.
It sounds like what needs to happen, is happening. Thanks for the detailed, clear description.
Yes, because we know the safety is improved on top of people's blood. Kinda poetic, but that what Petter (Mentour Pilot) said in his videos.
If Rolls-Royce can retrieve real-time telemetry from their engines in flight, why can’t the same be done with black boxes?
It wouldn't be done continuously. Aircraft have have had the option of transmitting "snapshots" of (limited) data for many, many years.... either by manual command, or automatically at certain points during the flight. Larger, bulk transmissions are often done in the vicinity of airports where VHF groundstations are used, rather than satellites.
@@ImperrfectStranger Or at least on certain events, like alarm popping up.
@@piotrd.4850 True. That, too.
I was thinking of RR and their Trent series. I seem to recall though that they have a dedicated team constantly monitoring the system and it's paid for by the aircraft owners who lease the engines. Thinking of the comment Petter made, the airlines would pass this onto the public through ticketing prices and it wouldn't sit well with the budget companies.
RR has this because they're paying for it. (and it is not remotely cheap.) As I recall, it's not real-time.
I’m a long term satellite comms professional - worked ten years for the leading provider of aircraft satcoms systems. The cost per Gb is not really a problem - $10/Gb for the satellite bandwidth. There’s a big cost in fitting the antennas and modems, maybe $1m if it’s not a line fit (ie built into the plane by Airbus or Boeing). Also while reliability is good, a typical satcom system would not provide data in a dive situation like the China Eastern crash as the ‘look angle’ of the antenna can only cope with a limited range of attitudes. Certainly it could be a useful addition though, especially where aircraft are being fitted with broadband satcoms for customer internet access, in which case the cost for the extra bandwidth needed for CVR and FDR would be marginal.
Can you explain how the antennas track the satellites please?
@@darrinito I am assuming the new method will have less antenna gain? I suppose being at 35000 ft means less atmospheric signal loss though so it would still work ok.
@@darrinito Ok thanks for the information. I work with radio telemetry systems but we don't use antenna systems as you describe. Off to Wikipedia to see if I can understand how they work. Thanks for the information.
@@geoffwoodgate7450 The key factor is the frequency band in use - Inmarsat and others provide service at L-band (~1.5GHz) and don’t require accurately steered antennas, but the bandwidth is low - too low for this type of application. For broadband connectivity we use Ku or Ka band (12Ghz, 29GHz) and at these frequencies you need quite a large aperture to get gain, plus ability to track the satellite to ~1 degree accuracy. This is tricky when the aircraft banks or dives as the radome housing the antenna on the top of the fuselage is quite low, restricting the antenna movements (whether mechanical or electronically steered). Look at the ‘hump’ on an aircraft and compare with what you see on a ship that doesn’t have aerodynamic constraints.
@@Watchyn_Yarwood Yes of course - you can get satellite broadband access on aircraft. But the cost of retro-fitting is high, compared to line-fit.
You're right in that they'll become supplemental system and not a replacement system. The broadcast system is going to be more succeptable to issues than the hard copy on the plane. Any extra data is welcome but they won't replace the "black box" system, just enhance it.
Fun Fact: Marine industry is also using “black boxes” named Voyage Data Recorders. Systems installed after the 1st of June 2014 are mandatory to have two different final recording media (that keep the same data). One fixed with a beacon and one that deploys automatically, called float free that starts transmitting it’s position at 406Mhz.
Another possible source for the name "Black Box": In engineering terms, a black box is a device or object we insert into a design per specification where we don't care about the inner details.
The first BB's inserted into a plane was just like that: Mount this box and hook up those cables. It doesn't contribute to the characteristics of the plane, so don't bother what it does.
Actually it was because originally they were actually black. They were changed to orange to facilitate finding them after a crash.
Black Box also often refers to a device or programming function where we know what it's supposed to output but have no idea how it does that. The inner workings are hidden from the user.
@@johnbower7452 I saw the video. It was only a possible explanation.
@@Seriously_Unserious Exactly.
While this is true, is the usage in engineering predating the one in aviation?
I was talking to my colleague about this just the other day! Thank you so much for covering this very interesting topic, and with such great timing.
4:15 in my opinion the term black box is a carry over from normal science/engineering usage where any piece of equipment that we don't need to understand in detail is referred to as a black box. A pilot and and almost any plane designer don't need to understand how the insides of the recorder works: so to them it's just a black box.
The only thing a pilot has to know is how to stop it overwriting the records immediately after a survivable but serious incident. Even that is not a feature of the black box itself: typically the pilot would simply pull the circuit breakers on landing in those cases
I'm old enough to remember a crash report on the TV children's programming from sometime in the sixties or seventies where the black box was shown being retrieved. The TV presenter explained that it was neither black nor a box. It was roughly a cylinder but with rounded ends. The shape made it more resilient in a crash and the bright orange colour (as he described it to viewers like me watching in black and white) made it easier to find. He suggested we should perhaps call it a red egg, but I don't think anyone in aviation ever used that term
The term black box has been used for any avionic equipment working in a sealed enclosure a long time before computers have been first invented.
Cool comment. Red egg lol.
Nope, the term was developed in parallel--if anything, the systems engineering usage was influenced by the aviation usage, but there's scant evidence for that either.
As Petter said, early data recorders were usually secret and secret components were often painted black. Literally a black box.
In circuit diagrams in the early 1940s, a heavy black box was used when a complex component wasn't relevant to the operation of an enclosing circuit. The diagramming conventions for this eventually changed, but the term stuck. It's possible this was influenced by the habit of painting secret gadgets black, but little evidence either way.
@@herestoyoudoc Vitold Belevitch puts the concept of black-boxes even earlier, attributing the explicit use of two-port networks as black boxes to Franz Breisig in 1921 and argues that 2-terminal components were implicitly treated as black-boxes before that.
Quoted from
en.m.wikipedia.org/wiki/Black_box
(Scroll to history section)
Avionics can be said to have started with IFF during WW2, or arguably when radio transceivers started to be used in the air, both of which came after 1921, of course. It didn't become regarded as a subject separate from general electronics till the fifties or maybe even the sixties. I stick with my original assertion that aviation borrowed a term that was already well established in scientific and engineering usage
@@trueriver1950 I'm not saying that it's impossible for the etymology to be the same, but the evidence is really scant and speculative. The simpler and high-probability explanation is that data/voice recorders were originally literally in boxes painted black and the name stuck.
Or, ultimately, with the additional background offered from Breisig, it could be the etymology is shared but by analogy to a camera and not for any other reason.
The US Navy P-3B had a data recorder that was installed on the tail. In an accident, the recorder detached from the aircraft and activated an onboard transponder. The data recorder was also designed to float if the accident occurred over water.
Exactly... You can still have a full data storage in the aircraft.
Loving the title: straight to the point telling us what to expect.. youtube nowadays is oversaturated with titles in the line of "why THIS plane failed" etc and it makes me mad. Your video was an insta-click because of that in combination with an interesting topic rather than a fishy clickbait.. much appreciated.
Given the number of incidents you have covered where the problem originated on a previous flight, how about downloading the data during turnover at the terminal? While it may not be helpful in all cases, it could still be helpful in some. This would also allow investigation teams to access even older data to see if anything was missed during previous flights.
My thoughts exactly, although 88 flight instruments' data could easily and cheaply be streamed live to detect a potential risk live on the on-going flight as well, not just future flights. But yeah, we now have ginormous AI machines that would just love to swallow up all this data and detect patterns across airplane models or flight conditions, maybe even specific pilots, if this information is used positively and productively (training) rather than retaliatory.
This is a damned good idea 👍
The black boxes should always be in the planes as a last resort fail safe option even if they do get this up an going at a high capacity. Redundancy is the key with planes and having more than one way of keeping track of all the data is going to play in the favor of them and the airlines and especially the people that rely on them every day.
Trey N said, "The black boxes should always be in the planes as a last resort fail safe option..."
------------------
You make the point well-- too many critical systems succumb to a single point of failure. Likewise, a broadcast stream of data always benefits from another format-- an onboard recorder. Post-crash data is usually bought at great price, and having an alternate system is a cost-effective remedy.
another possible benefit from contiuously uploading to the cloud would be that someone at corporate could look and find and suggest solutions to issues.
I'd love to know why the regulations still only require 2 hours of cockpit voice recordings. Especially when ocean going cargo ships keep the entire weeks long trip
I'm not sure how often this happens, but I've seen a couple of cases where the black boxes lost power and stopped recording early. They should hand an independent battery backup. It would be minimal expense and ensure up to the last second data that could be significant in stopping future incidents.
It is insuficient, when black box work, but data feed is down due to powerloss of plane. You know. Black box lost power for a reason.
essentially all that would be needed to be transferred to sattelites would be it's position, because in that case they would be able to recover it's location so we don't have unsolved mysteries like MH370
And that’s what ADS-B is already doing.
@@MentourNow Indeed and with satellite based ADS-B receivers we get global coverage. What it won’t stop though is a rogue pilot pulling the breaker on the transponder, or on a satellite internet system before deliberately losing the aircraft, as may have happened in China Eastern and MH370. The flight crew always need that option in case of fire or malfunction in the electronics
This, exactly this! Everyone keeps saying it's too much data for every plane to contanstly transmit black box information. Commercial flights that have 100+ seats should be made to transmit latitude and longitude every 2 minutes. Is that really too much data to deal with? Make the system on by default, you can't switch it off. It'll be much eaiser to find black boxes.
@@MentourNow but ADS-B does not have global coverage yet?
@@AnzacPT Not all aircraft are required to operate ads/b out
Not all airspace requires ads/b.
ads/b is neither authenticated or encrypted.
imo, there's a lot of stuff that could be done with compression, and intermediate data packages in case of value spikes while maintaining a lower update rate outside that for regular purposes. for a "not a replacement system", but a complimentary system that provides at least some data for these situations where the data recorders are not recovered, or damaged. especially combined with panic buttons that both software and pilots can trigger, to start transmitting more detailed data, this seems like a very good addition to existing systems.
it could also provide ground crews with additional tools to support aircraft that have not yet crashed, but are experiencing problems - hopefully preventing the crash all together, by allowing a team of engineers on the ground to double-check the status of the aircraft systems and helping the flight crew troubleshoot the aircraft. who could even have access to a whole different level of tools to look at the data - like, if there's a suspicion of a problem with a specific engine, they could with just a few clicks see the performance graphs of that engine during the entire flight, and even get comparison data of the previous flights to form a hypothesis of what's going on and what steps could be taken. like "yeah, i can see that the number two engine has been running x degrees hotter than the number 1 engine, and has been holding that level steady for the last y minutes. there's also been a drop of hydraulic pressure back then, but not since"
it could also allow for a more in-depth tracking of the time some systems have been active, possibly leading to improved maintenance intervals - like, on this aircraft, the flaps have been deployed for this many minutes at speeds between x and y, exceeding the "normal" rate of their deployment at these speeds by some margin, and increasing the load on the involved systems, so they should actually be checked a bit earlier than the regular maintenance interval.
it'd also allow for some external validity checks that might not be feasible to implement on each aircraft - like, for example, it could gather the metar data from all airports in real time, and check that aircraft at or below certain altitudes near the monitored airports have an altimeter setting that's within some margin of the actual current meteorological conditions, and if not it could alert a person on the ground to the possible mismatch, providing the live data from aircraft and ground plus the latest metar update, plus maybe information from other nearby aircraft for comparison, and that person then can determine if they might need to contact the aircraft to alert the crew to a problem. i'd imagine that there's a lot of checks that could be run on the ground side that should get filtered through a trained human before alerting/distracting the pilots in the air. things that can be easily expanded upon without requiring updates to all kinds of onboard systems, or being flight certified, since they don't run on an aircraft system and get double-checked in a non-critical environment before being brought to the attention of the pilots.
A lot of your (good) ideas are already implemented as optimizing of the maintanance and in result avoiding unplanned availability of aircrafts saves airlines a lot of money.
Your idea of having a person on the ground who is looking for possible mismatch in the data of an aircraft would be very expensive as it adds a lot of headcount.
I agree that a lot of checks could be done - but soon as you aware of that should be checked to identifiy possible mismatches it is easier to let a computer do all these checkings.
A lot of accidents are caused by wrong sensor data confusing the pilots which data they should trust on which is understandable in a stress situation where you get mutually contradicting data.
I think here additional software could help the pilots a lot by telling them how reliable each data source is.
For example if one altimeter tells you (by defect) that your are flying at 30000 feet while the other tells you that you are on 3000 feet the software could use additional data from GPS to determine which of the both sensors most probably is wrong. Even the outside temperate could help in this decision - at 30000 feet is is always much colder than on 3000 feet.
Engineers can spend a lot of time to think about and define such realiabilty calculations for each sensor using data from other sensors and a software on board could use these calculations even in the most stressful situations to show them which instruments they should trust and which not in case they are in doubt.
@@elkeospert9188 thanks for your reply!
> "I agree that a lot of checks could be done - but soon as you aware of that should be checked to identifiy possible mismatches it is easier to let a computer do all these checkings."
i meant that, you have computers run those checks automatically, but do not automatically send information back to the pilots. rather, the information that gets flagged by the computers then gets double-checked by qualified people (who only check those flagged data points and not monitor everything all the time) before the pilots get notified.
the pilots might be in a very stressful situation, and i'd rather not have automated systems on the ground add to their workload, but have a trained human being assess the situation and make a judgement call there
> "A lot of your (good) ideas are already implemented"
glad to hear that they've had thought about that stuff too... though i'm not really surprised, tons of good people working on improving those things :)
I have to question your data volume assumptions. 88 parameters using 2 bytes per sample of data updating 10 times per second would be 14 kbps of bandwidth. That's old timey modem speeds. Taking audio from 3 sources, moderately compressed, it would add around 480 kbps, totaling 500 kbps. That is such a small amount of data, when these airplanes already offer internet services in flight.
By all means, keep the black box, you should always have a physical backup you could access in case something goes wrong. But, you'd save a lot of cost and time if you could always access the black box data (even in a limited form) instantly after an incident, and never have to worry about spending the time and money on a search. Sure it is rare for an incident where the black box cannot be recovered, but not at all rare for an incident where it is badly damaged, or cannot be immediately found.
I also mentioned this point in my comment. In its principle I agree with you - it’s not a lot of data.
I would go on and argue that it’s actually much lower than 500kbps. Your calculation is correct when the audio is analog and uncompressed. But a digitalized and compressed audio file (which makes it worth transferring over data) can go as low as ~10kbps without losing much quality in scenarios like a cockpit.
@@Marcus-cc72 how efficient modern codecs are is insane. If all you need is enough quality to say hear the pilots speaking you could go as low as 6kbps or perhaps even less and it still be understandable. (You'd lose a lot of quality in things like background noise which of course could be important in an investigation, but as a backup method if the blackbox can't be found, it would be fine).
I'd guess with compression you could constantly broadcast a handful of key parameters and voice data with the same or less amount of bandwidth as a 56k modem.
My suggestion for numbers.
128 parameters at 16 bit resolution, 1 sample per second = 2 kbps .
Voice only needs telephone quality - 3.1 kHz bandwidth.
Min sample = 2*bw = 6.2 kbaud At 8 bit quality = 49.6 kbps per channel.
Allow 8 sound channels = 400 kbps
We're all getting the job done with 0.5 Mbps data rate - 1/40,000th of what our host suggested.
@@elijahwatson8119 right with you here. I’m not the best informed in terms of sound file encoding but I’d expect it to be around that ballpark too. Plus CVR quality is already pretty nasty anyway.
@@Marcus-cc72 I'll note that CVR is typically raw WAV format at 16bit / 44.1KHz (CD "quality") and ask if that changes your conclusions at all.
I worked on a prototype study back in 1998 which looked at the feasibility of doing live data transfer for the black boxes. At the time I was calculated that this would only become economic when the cost of transfer fell to below $1 per gigabyte of data. I think we will probably be there in about five years.
How much did it cost then and how much does it cost now?
@@ChrisD__ At the time it was about $20 a gigabyte. Not sure what it is now as I am no longer in the industry. The last price I heard, which was in 2010 was $8 a gigabyte.
@@nigelgordon Why do you need gigabytes? You can do this with KB. How much data do GPS coordinate use?
@@mortgageapprovals8933 You are dealing with a lot more than GPS data, which is fairly small. On the study I was involved with we had to record over 7000 data points, each of which was a 64-bit data value. The requirement was that we sent the data point identification, the time of the data read, the 64-bit data value and a checksum that is 18 bytes of data per data point or 144 bits. Each data set had to be read every 10th of a second. Add to that the cockpit voice recorder data get a lot of data having to be upstreamed. With flight data about 4.5 gigabytes per hour and cockpit and cabin voice recording coming in at about 1 gigabyte an hour (multiple pickups being recorded) the test ended up running at just under 6 gigabytes per hour.
SpaceX’s Starlink will change this forever
We can never replace the blackbox but i think, while having the blackbox i think sending extra copy to the server during abnormalities would help if the blackbox doesn't survive, even the litle data sent out before a crash can give an idea closer to solving a tragedy, would give a general idea.
Why not just send the voice/data recorder data through ground/satellite with a lower resolution? (e.g. 128kbps audio quality, and less frequent parameter sampling, which would require a data transfer rate of less than 1Mbps). I'm not suggesting this is as a replacement to the 2 recorders, but as a complementary measure. The cost of this surely won't be prohibitive (both to implement and to pay for data transfer). This could certainly help in those rare cases like China Eastern 5735 and MH370, and is better than nothing.
Somebody else posted in another comment that it wouldn't help. In a dive like 5735, the plane to satelite connection would be lost because the transmitter/receiver would no longer be pointing at the satelite, simply due to the orientation of the plane.
The broadcasting cost is one issue. The installation/design/certification cost is another. You still have to justify the cost for rare suicide events. Not mentioned in the video are the human issues. Not all pilots like "big brother" listening to them. I don't know if this is still the case, but pilots used to have the option of deleting the audio at the end of a flight (when the park brake had been set).
Telephony quality only needs 7k bandwidth to carry something that is capable of being understood.
@@trueriver1950 doesnt work well on the kind of audio in question.
@@danielch6662 Services like starlink use phased array antennas (meaning the "beam" is steered without any physical movement), and if correctly configured and with sufficient satellite coverage, should still be able to communicate with satellites even when vertical.
About uploading data to the cloud: Sending data after declaring pan pan or mayday... (Or punching in some code by cabin crew) would reduce the amount of data sent under normal circumstances, but in an actual emergency scenario it should be helpful. If it is sent in a form that enables training pilots, engineers. Using high amount of bandwidth and storing a lot of data ONLY when needed would be still affordable.
Even if we would allow the aircraft to initiate this data transfer on its own, in reacting to TCAS and EGPWS warning, or maybe in reaction to some fault, that would enable airlines and other operators to start reacting to emergencies before the pilots even have a chance to make a mayday call.
Of course these networks can be made in a way that they would allow 2 way communication allowing to help the pilots with some visual information.
I would prefer if the data broadcasted would enable people to see the developing emergency in a simulator. If simulator training staff, who should routinely deal with various (simulated emergencies) can quickly understand the developing scenario and can communicate with the rest of the company to help the pilots as needed, then this communication system would do MUCH more than replacing the black box.
Replicating an emergency situation in a simulator on the ground, is a great idea! The fresh mind of the simulator pilots would certainly help where the aircraft pilots are mesmerised by a stuck autotrottle, or being confronted with the 737max nose dive, where it was necessary to take the unnatural decision of decreasing the speed of a falling aircraft in order to regain authority on the control surfaces in manual mode.
Well said.
Except in the vast oceanic expanses, there isnt enough data to send a text - let alone what you suggest. Unfortunately, too many people carrying a smart phone think the rest of the planet has the same coverage - it does't and its not even remotely close.
@@grahamo22 Except for your comment with the implied insults forgot at least 4 different problems.
First of all, your GSM phone is a land based technology, so of course you can't send anything where there are no towers... But it has nothing to do with data or bandwidth, but with coverage of a single technology
But there are global satellite based communication systems, and even a few decades ago they were used even for live TV.
Your lack of knowledge and technical understanding wouldn't be bad if you wouldn't throw implied insults only backed up by your lack of knowledge.
Second: Existence of both long range and satellite based communication were acknowledged on the channel, when you pretend that they don't exist to throw implied insults... That is already a bad idea.
Anyone interested in civil aviation would see that people often review on board WIFI and it is easy to be aware of the feasibility of some internet connectivity that can be shared by a lot of devices. So your offensive comment didn't only ignore old information about different networks of technology, but also common knowledge about aviation and on board services.
Third: The video explained that it is easy to find the black boxes too because most emergencies around airports, also shorter flights over land have worse statistics, etc. so your argument about oceans is totally irrelevant in majority of cases.
A dick comment that implies that others are dumb and ignorant, while shows only your ignorance is usually not a good way to engage in a civilized conversation.
Fourth: We see more and more satellites for different purposes each and every year. Even if we ignore existing technologies mentioned, that wouldn't prevent deploying new satellites. ADS-B Satellites are also somewhat new and mentioned in the video, and new satellites are often made for new purposes.
So while you might be unaware of things like Inmarsat: www.inmarsat.com/
And you probably don't know that the Satellite Data Unit of Malaysia Airlines Flight 370 was in contact with Perth, how they used the "Classic Areo" plan at the provider (yes, the name implies, newer and better plans) and that Inmarsat S EAN is EAN for "European Aviation Network" Or you don't know Globalstar either, as they aren't that well known services, Iridium was in the news often, etc...
But these companies aren't only used by aviation or news industry, but they often provide a communication architecture for ships also and can help with providing internet for tourists on cruise ships. The need and existence of these systems are somewhat common knowledge, but even if you don't know about the existence of these systems, you don't know the global satellite based internet service provided by Musk (and used in Ukraine by their army too) so you managed to ignore all the news and common knowledge, and you lack the upbringing to know that you shouldn't paint others as ignorant when you lack the basic knowledge about this field, it would be still common sense to expect development of (more, better) satellite based communication services.
@@grahamo22 you can connect via satellite ANYWHERE in the world, but right now it is expensive for commercial use... How do you think a plane can provide wifi connection to its passengers if not connected to a satellite?
I don't think it would ever be reason enough to get rid of flight/cockpit data recorders - even if we got to a place where it was possible to stream all the data reliably in real-time. Why would anyone ever get rid of that belt and braces backup? I like the idea of the bursts of data if something feels off or that the pilot can manually send, and that it records/sends x amount of data before that point. It reminds me of back in the day when dashcams only saved the previous 30-60 seconds of footage before a crash or whatnot. It seems like a really good compromise considering the current safety and data transferability.
Redundancy is the name of the game when it comes to airlines, so I think that the data burst idea in addition to the physical black boxes is an excellent one. There is always a chance that the black boxes will not be readable thanks to damage, and there is always a chance that the data will be corrupted, so they can cover for each other.
You can compress data, also you can also prioritize various data streams from instruments. So it doesn’t have to be as high resolution as whats typical in the actual black box recording
Still doesn't solve the problems that arise with a failing connection. (be it partial or full failure)
@@TheExileFox I make systems that tolerate intermittent connectivity all the time. It sends data, gets an acknowledgment of what was received. Keeps trying on the segments that didn’t make it through. You would still have a high resolution recording on the planes black box. What’s uploaded to the cloud would be something skeletal in comparison
a note on black boxes: "black box" is also used to refer to a system that you use without understanding exactly how it works.
That was my assumption for where the name came from.
I would argue it is used for the systems where you do not care how they work.
@@NetAndyCz those are included in the category. for most people a computer is black box technology. for some, just about everything is. addendum: it is also applied to stuff with no user serviceable parts inside.
Mentour- my question is why are there no cameras on airplanes? Today cars have them to reverse, to aid with side view blind spots etc. When you go into any number of public places there are cameras watching your every move for security reasons like banks, casinos, hotels, etc. So many issues even after finding the black boxes leave questions unanswered as to what "really" happened. Cameras should be on board in almost every spot possible inside and facing outside of the plane. If a captain needs to check if a engine is actually on fire or leaking fuel or is landing gear locked/down he/she should be able to flick a switch and see it on camera. If a passenger is causing an issue a captain or even in the flight attendant galley should be able to see it on a monitor at any time. All this should record to the black box as well. The cost of this technology is negligent and the size of cameras now would be invisible to the public but so helpful in every way from day to day use, safety and trouble shooting after the fact in case of accident. Mr airplane executive, we can offer you our new A-320 plane standard features at 100million$ or do you want the upgraded version with cameras/monitoring with upgraded blackbox for 20000$ more? No brainer to implement!
Many countries have privacy rules that prevent recording of employees or customers without justifying the need. A brief, voice only recording accessed only AFTER an accident is one thing, but the legal quagmire of multiple national laws involved, for the extremely limited potential benefit, make the idea of random video surveillance something that probably just isn't worth it.
Mark is right - and it's not just countries. The GDPR is quite simple in this: unless you have a very good reason, any form of camera placement is absolutely prohibited. As an example: if you have a supermarket and you want to put up a camera in the stockroom, you can only do that if you can show that products are disappearing. Besides, how would you as a passenger feel if your suggestion was implemented - including the lavatories?
could implement a camera that start recording visual and voice commands whenever 7700 is squawked, which could be compressed even streamed, so that if for any instance all options appear to be have been exhausted , more Professionals on the ground could offer advice in real time.
I really think this is a great idea. I don't see any privacy issues because the recordings are deleted as soon as the flight is completed. Cell phone cameras are now so small, so good, and use so little power at the cost of about $20 or so. This would have been very valuable in the 9-11 hijackings and the china crash. Like the cockpit voice recorder, the actual footage would never be released. Only a description of the events. I'm pretty sure the NTSB has been pushing for this for many years and it's being fought by the pilots union.
I imagine making the cameras weather proof and durable would be the greater cost. Airplanes face more of the harshest weather conditions than cars.
For me as a computer scientist this was a question I had for quite a long time. Sure, live streaming of data isn't a good solution for the reasons you mentioned but that idea of sending data only event-related sounds like something that should at least technically be absolutely possible. And with the costs going down quite fast, it'll be not long when it's be cheap enough to maybe be enforced by authorities one day.
20 GB seems quite a lot.
If 100 parameters are sampled as single precision floats at 1 kHz that would be 1.44 GB/hour.
And another 1.44 GB/80 min for 4 channels of uncompressed reasonably high quality audio. (2 CDs)
@Foobar You forget that it's probably not only random float values thrown into the blackbox (or into the streaming to stay in that idea) but rather raw flight data along with tons of metadata and some stuff we as amateurs don't see. So you cannot take e.g. valve position as one raw float value but you have to consider stuff like "which valve is it", "actual flow rate", "normal flow rate at that valve position and at this phase in flight" and maybe much more. And all that needs to be packed into a processable format which in itself also produces some data overhead. And this is not considering networking and protocol overheads and so on when you do live streaming of flight data.
In the end all that may be absolutely possible technically. But as long as lost blackboxes are that rare and live streaming blackbox data isn't mandatory, it comes down to cost and value. And even if safety is (or should be) number one priority, it's not a big secret that airlines try to safe money whenever possible as long as it's not impeding safety.
@@Sinned1208 Yes, but it doesn't need to repeat the metadata millions of times, that would be an even more terribly inefficient format. There just has to be a lot more than just the 88 flight parameters for there to be 20 GB of flight data.
In order to receive good amounts of valid data, you need bigger bitrate for reliability and stability. Ping also plays an important role in data streaming. Most of us tech junkies relied on fiber optic and transoceanic internet connections in order to get stable service while airplanes are completely wireless while airborne . And trust me, live broadcast TV expend hundreds of gigabytes of bandwidth everyday just to stream video and audio. Flight data is way too complex for another computer to compress it then transmit it.
i just cant go to sleep without listening to this man , i love his explanations
While black (or orange :p) boxes won't be eliminated anytime soon, it would be nice to keep improving them as technology is advancing: make them record more data (how about 24h of CVR for example), make them more durable, easier to find, and easier to handle after recovery.
If airlines start using Starlink for their passenger data anyway, could they potentially use a data channel on it for flight data? Maybe not the full amount that the black box would record (especially not the voice recordings), but basic telemetry and errors/warnings maybe?
Might be a fun project for your discord to chat about what would and wouldn't be useful to send that way
Well, at the (not too distant) point where Starlink will be providing fast internet access for everyone on a plane (i.e. half of the passengers being able to stream videos at 1080p or more), the 20GB of telemetry and voice on a long haul flight will be peanuts compared to the data transmitted purely for entertainment.
But even then, the satellite connection will only work as long as the plane can maintain some sort of stable attitude. For example if it was flipped upside down, the satellite transmitter would point in the completely opposite direction. The pivot mount and phased array antenna can only adjust/track for a certain number of degrees, not 360° in all axis. And probably not fast enough to keep up with extreme maneuvers either (especially roll).
So keeping the black boxes will always make sense. But as more and more data is being streamed even just for the benefits provided to fleet management and technical maintenance, we will most likely see almost zero cases of completely "lost" aircraft, because there will be some data available up to the point where the aircraft enters such a violent attitude that impact/breakup is likely only seconds away. (Well, unless we get another Air Astana 1388...)
The amount of telemetry in a modern aircraft is astounding compared to the rather minuscule amount of data required for an audio stream. Modern codecs like Opus can encode voice at 56 kbps, which can be achieved even on an antiquated dial-up modem. And that's for real-time communication. If you don't care about a 100ms delay, you can do even better. A 1TB microSD card can store 4.5 years of 24/7 audio.
Like said in the video: Mostly you don't need that much data. Sending your location and some basic telemetry data via ADS-B and sending a burst of information when the system detects an anomaly sounds (like said in the video) sound more practical in my ears. I'm no expert at all, but transmitting all this mostly unneeded data seems like a huge waste of bandwidth. Also there is another aspect: I can imagine that pilots could feel really uncomfortable when all of their (private) talks in the cockpit would be transmitted and observed in real time. There is one rule in data collection: Collected data will be used and often it will be used for more then the intended purpose.
@@marc-andreservant201 Despite their name cockpit voice recorders record a lot more than just voices. If you use a codec designed for voice comms you will lose a lot of the ambient noise which can give clues what happened. The reason they are uncompressed is because you want to hear everything, at all frequency ranges, when studied at a forensic level. The moment you add any form of compression you can no longer trust what you're hearing is what actually happened.
Starlink is at least for now just more vaporeware it seems. I doubt it would be remotely useful in this scenario.
40 or more years ago, the German Luftwaffe Tornado IDS aircraft was fitted with an ejectable (and if I remember correctly) combined ADR/CVR. The aerofoil-shaped device was fitted on the spine behind the cockpit and was ejected from the aircraft in the event of the wingtip or nose mounted crash-switches operating. The device was ejected into the airflow to clear the airframe and land away from the crashed aircraft. It was designed to float and contained a battery-powered transmitter to allow easy location after ejection. This technology has existed for years, so with the immense cost of trying to locate a sunken aircraft and understand why it was downed with implications for the remainder of the fleet, the system could easily be fitted to aircraft either during build or retrofitted. Yes, there's a cost, but economies of scale would apply and it just needs legislation to enable it.
Modern technology would allow for more data storage with multiplexing and data-compression techniques plus rugged storage devices would survive fire, impact and water immersion. Basic ADS-B with positional data would keep track of the airframe if receiving stations could be provided to increase coverage, maybe with satellite or balloon systems.
Adoption of such a system just needs the industry to realise that it costs a damn site more money to locate and find out why an aircraft has crashed or disappeared than it does to proactively pre-empt the problem. After all, aviation safety is no accident!
Great Job mentour and the team! The quality of your videos is just way above the bar, and it just keeps getting better day by day!
I personally just want to say being a long time subscriber, i am loving these good old style technical deep dives like you used to do back in the day, And i do hope you continue doing them!
And honestly just my opinion of course, but i think with how good the production quality on these videos are, you really could consider making long form documentaries on various aspects of aviation.
You know like a proper truly deep dive documentary going into the history of it, and the backstory, with wonderful graphics and visuals.
Similar to what you might see on some of these other aviation youtube channels like Mustard, and paper skies, just to name a few, who make these really high quality documentaries on specific aircrafts.
I really think you could take this a step further, not just talking about aviation accidents, but full documentaries on various iconic aircrafts, and the development of technical systems like what you talked about here about the black boxes.
Just my two cents, i think you could really become THE channel for all things Aviation with how well researched and detailed these videos are. And i must say you truly are a wonderful explainer! The detail and objective and thoughtful analysis you do is beyond anything i have seen anywhere on youtube!
Anyway those are just a few wild ideas i had, Love all your content as always! And looking forward to seeing this channel grow to new heights!
On the contrary, content still great, but presentation has gone way over the top with unnecessary and distracting music and video effects.
@@kslingsby The looped Musak has gotten way out of control on many RUclips videos, at least for my taste.
It's quite simple. You just include a simple api to parse the uploads in stream segments just like exchanging ffmpeg or vlc inputs and outputs into groups of nodes. When one stutters, the other 8care there to determine the missing content, patch it, and match the input keys. Keys match = content successfully streamed into analyzable binaries. Binaries incredibly lightweight, non radio/transmission harming anymore than in flight wifi or carrier movie and show streaming on the client facing server. You could even take a playbook from Silicon Valley, and stream parts to the carrier app itself, and offload and uncompress the secure content on landing; parse it off to db server, remove from cache of app, and bam done. All within 2-5 seconds, if not under 10 milliseconds, and all under 150kb on the longest flight you can physically book (28 hours of black box data under 150kb)
The tech is there. The investment in competent devs at the carriers in line with those at the FAA is the issue. The devs at Boeing and hardware/software backends are not the problem.
@@anon_148 How do you think carriers and streaming apps offload a 50-300GB movie or series into compressed chunks at 100-300% less size? They utilize timed compression layers. Slowly, processing enough data (buffer) a little bit past what the user needs, and nothing more, while slowly placing checkmarks on each chunk, so that no data is lost. (skipped) In this case, it could be on a pre-timed program throughout the flight. The entire Netflix library is hundreds of thousands of GB's. All at the top of your fingers, pre-playing away, utilizing various techniques to deploy that content to you at a moments notice. Placing the loads in to chunked client-server-serverless nodes allows that process to be lightweight and fast.
The ability of the pilot to transmit a portion of the data (e.g. latest 30 mins) could be useful for troubleshooting issues from the ground in certain situations
It should be automatic, every alarm should be sent to a technical support center, so they can help troubleshoot in flight.
Fascinating - I really enjoy these technical insights and hope you keep producing them. Thank you.
The idea about having a "burst backup" in case of any failure makes sense. Either way the system should have offline storage buffer in case the network is down, so a rolling backup of the offline buffer would be the way to go even if the entire blackbox contents were backed up when satellite internet gets cheaper. So that is a good first step.
Also, lots of people underestimate how expensive and limited bandwidth satellite internet is. Transmitting gigabytes of data on the regular via satellite is still *very* pricy, for example on ships you usually buffer non-critical data during voyage to be transmitted via cheap networking in port.
This will hopefully improve in the near future (with initiatives like Starlink).
Keep the physical box in the plane, but also stream it to the ground. Instant backup if needed plus less likely to lose the data
It kinda feels as if you didn't watch the video :)
@Alfred Weber he kind of did. It's a cost vs gain vs possible viable alternatives (manual/semiauto data bursts during abnormal situations) argument, which he did.
I just can’t agree. I wear a seatbelt despite the rare chance of an accident. But I can think of a lot of reasons why a pilot wouldn’t want idle chatter being uploaded to an easily accessible medium 🤷🏻🤨
As a person who monitored lots of sensors via modem, I say it's possible. As people who played online games most of you will agree. Because:
20G/2h ~= 166M/min which looks significant. And it's uncompressed (or only lightly compressed because it makes post-mortem decoding harder, but this is unconfirmed!). BUT you don't have to stream audio! It is already transmitted over analog radio! I'd say transmitting realtime critical sensor metrics is more than enough in emergency situations.
To make a fair analogy: a multiplayer game (which is sensitive to time lag) transmits tons of parameters of any kind from boolean on/off flags to floating-point coordinates, and we played FPShooters on a dial-up!
It may also help people on-gound to see the control panel and help the crew right away if possible.
Petter, you did mention in one of your recent videos that aviation rules are written in blood. Just when regulators think they've seen everything, something new and unexpected comes up that forces the hand of the regulators once again. We don't have the CVR and FDR of flight MH370 but there are strong pointers to it being a pilot suicide- one of the most difficult to counter problems in aviation simply because there are little or no warning signs. In the case of the flight MH370, the existence of a simulator flight plan on the computer of the Captain who flew MH370 which ended up in the Indian Ocean that matches the estimated flight path by means of doppler satellite pings, the recent revelation that the plane went into a holding pattern at the end of the flight etc show a pilot fully conscious of his actions- trying to redress some unaddressed grievance in the most extreme manner by plunging a plane full of passengers and himself deep into some of the most remote territory on earth. I'm all for the old technology but what about the future? Mass Gun shootings were once quite rare in the United States but these days they're happening with alarming regularity. If pilot suicides become more frequent, we're going to see more calls for real time uplink of data, no matter what the cost. If I were the regulator, I would mandate live streaming atleast for flights over open water because recovery from ocean depths is often quite time consuming and very expensive.
CVR and 1 second snapshots of the FDR or QAR data should make for a decent first generation.
the problem with the pilot is he had over 30k hours i believe, he had a family, he knew aviation inside and out, more than you and me combined. i dont think a pilot like that would consider harming his passengers for political reasons. The holding pattern indicates that too, because if he was in a holding pattern deliberately then that means he was trying to do two things at once: either defend himself from a crew member or a passenger, or maybe it was a failure in the aircraft that they just couldn’t figure out.
You know that youtube live streamings have delays, or how about live news reporting are actually prescripted because of the 10 seconds delay. Streaming videos is already a bandwidth hog costly. And you're talking about flight data that we don't know how big it is. Aircraft manufacturers know that airliners know that it would become more costly to operate such feature and therefore even the tickets could not cover the cost of one operation.
Hi, love the content 🙂, just wanted to mention that the background music is quite a lot distracting. I know, many people do not notice it and dont mind, but it is one more reason not to use it at all. In some short sequence for dramatic effect background music is great, but through the whole 17 minute video it is too much. Thanks for the great content anyways!
I came to comment to say exactly the same thing. The music was so distracting that I lost track of what was going on in the video more than once.
I don't hear any...?
Agreed. I have hearing problems and found it interfered with the narration.
@@RyanEmmett I absolutely hated the repetitive elevator or worse music
I agree. But even worse, the sudden sound effects are disturbing as well without adding any value at all. Same for some flashy transitions and these strange glitches in between. I rather get the impression that someone is playing around with the cool effects of his video editing software.
With things like Starlink, I imagine it will be far easier in the near future for aircraft to stream information, though it should be clear that anything like this should be _in addition_ to a physical black box, not instead of. And as other commenters have said, having e.g. 1TB of solid state storage, in addition to what they already have, would be a cheap addition.
My day job is IT Petter and I agree with you completely for all the same reasons. An uplink is only going to be useful for those rare occasions where the FDR and CDR are destroyed.
Your informative video on the black boxes clears up many discussions on replacing these boxes. In the conclusion of this video the short blast theory, sending an immediate help signal, needed is the best idea. For all the reason given the short blast ability to send immediate call for assistance would keep those monitoring the continuous volume of data from being desensitized from information overload.😮
The cost of streaming black box data via satellite should be contrasted with
a) the cost of searching lost black boxes from crashed aircraft and
b) the cost of the data that is lost in case the black boxes aren't recovered.
Especially in case b) the lost data is almost invaluable so it seems like the expense would be pretty easy to justify. It is rare for the black boxes to remain undiscovered but we don't know what caused MH370 so it could happen again any time.
I'd consider data security. Faking data inside the black box is close to impossible and online black-box service will suffer from attacks.
@@809Kocur Online attacks aren't going to do much to affect the transmission of black box data IMO. It's just send and store
If you consider how many flights there are every day, and how many crashes where we don't find the black box, I think that there is absolutely no way that the cost of finding the boxes comes even close to the infrastructure costs.
@@809Kocur security I would think is essential but it's implementation is trivial
@@jsfyxzuf117 if, for instance, the entirety of Sikorsky's in-flight data were to be duplicated to Bell in real time the industrial espionage would be worth millions, then there's the possibility of diverting & corrupting data in transit, for example adding a tail-rotor vibration to every example of a rival's model in turn so that the entire marque is grounded till the fault is found.
The scope for misdemeanor is endless, security must be very carefully considered & implemented in the initial concept.
20GB transfer over the satellite seems like a tiny cost compared to a cost of a single flight. Especially if it's bought b2b in massive volumes and internet connection is becoming a standard on planes anyways.
But even that emergency burst can be virtually unlimited in length - start streaming current data + best effort stream older data backwards as far as it goes until the plane lands safely or transfer is disabled by a pilot (last part is questionable considering the china accident).
In a long term if this becomes reliable enough it can even replace the black boxes as we know them and be a net saving.
Sounds like you've never paid for satellite time (or data.) I'm not talking about Starlink or HughesNet. Go price Iridium and BGAN. (still have a BGAN unit. that's a SIM worth it's weight in anti-matter.)
And go tell us how many hours it would take to send that amount of data when in the Pacific or Atlantic. Then have a big slice of humble pie.
Yeah, good luck convincing US carriers to spend more on this technology. They rather find unique ways to cram more people into the cabin and operate at the bare minimum service levels than worry about the "just in case" scenarios.
Cough* cough*, bitrates. Cough* cough* ping. Cough* bandwidth cough*
Absolutely any system must only supplement the black boxes. Any system needs power yes but to be truly useful it must be isolated from the crew so that it cannot be turned off.
As a minimum an intermittent location, altitude, course broadcast would be useful as a defence against rogue aircrew, or hijack or even just to find the thing. An early notice of a deviation from flight plan might also help in emergency actions (eg rescue).
The thing you are asking for is called ADS-B and is already mandatory on commercial flights
Also black boxes now come with their own power supply able to provide electricity exclusively for them in the case of a loss of power (already extremely unlikely since the aircraft has one generator per engine, another one for the APU, the batteries and then the Air Driven Generator, a windmill that automatically extends when all of the other power sources are out)
Source : I'm an aircraft maintenance engineer specialized in avionics
@@sixtealbisetti2480 So where is MH370?
@@emmabird9745 ADS-B exists in its current form precisely because of MH370 🙂
A “Black Box” is a common term in the electronics industry.
Whenever you need to use something you are not allowed to or don’t need to know how it works, it’s considered as a “Black Box”.
This could be a module designed and built by a third party which keeps the inner workings a company secret or a module which can have different functionality and or is based on different technology independently of your design.
You have a set of specifications of incoming and outgoing signals you need to fulfill in your design.
An example would be the CI (Common interface) slot on a TV. This was used for Pay TV before internet streaming. For the TV manufacturer, the card reader the customer can plug in is considered as a “Black Box”. The manufacturer of the TV doesn’t need to know which kind of Pay TV someone would use with the TV. The manufacturer of the Pay TV system on the other hand would consider the entire TV as a “Black Box”.
When you are using a “Black Box” in your design, you need to make sure that all your signals towards the Black Box are within the specifications given by the manufacturer of the Black Box. You can’t check your design by observing the Black Box if it is functioning as expected to validate your signals. If the design of the Black Box is changed and your signals are not within the specs, the Black Box might malfunction or stop working.
Test equipment and monitoring equipment is commonly considered as a “Black Box” by the developer of a circuit.
It is natural that the avionics developer will consider the FDR as a Black Box. All the avionics developer needs is the specifications of the dimensions, power consumption, type of plugs, pinout of the plugs and the specs of the signals going into the FDR. How it works doesn’t matter. Knowing the true specs of what the FDR could handle can even be dangerous by being able to cut corners in the design. So it is natural that the airplane manufacturer refers to the FDR as a “Black Box”.
Also there are two kinds of quality management. “Black Box Testing” and “White Box Testing”.
“White Box Testing” is knowing how the component (Hardware and/or software) is made and how it works. Then you think up ways to fool it or make it fail within certain parameters. In “Black Box Testing”, you don’t know how it works so you can only randomly apply data or signals and observe its reaction.
So my best guess is that crash investigators and/or the airplane manufacturer referred to the FDR as a “Black box” to the press because this is what they consider it to be. Something they don’t need to make the airplane work but they need to feed it specific signals and power in a certain way. How it works and stores the data and if it is certified or not is the problem of the FDR manufacturer.
As others have stated and talked about, there is no expectation that this would be a replacement for black boxes, and the figure of the data is in no means out of the realm of possibility. Being a software engineer (as one of the top comments is as well), I can agree that the amount of data here is actually near nothing in reality, and this isn't a system that needs to always be on.
If we look at it as a system that is not a replacement, but an addition to the black boxes, we can omit or compress some data. Things such as air speed may be measured in 32 to 64 bit values, however 16 bit values here may be completely plenty for instance for its purpose, hell, we can send the difference in data with periodic full values to save bandwidth as well. This is also a system that can *conditionally* send data if any system has detected an issue, for instance if any warning is detected, out of normal vibrations, ect., the last X seconds, or minutes of data can additionally be sent for getting the conditions behind such an issue.
Additionally, traffic is not as expensive as *storage* of data. Regardless of method (unless you're doing something crazy like sending data to and from the moon). This is a cost that can be mitigated by simply deleting data after as long as a week, or having the dataset reduced based on select parameters (master caution may retain data for longer for instance).
All in all, it is a shame that this system has not yet been put into service, there are no excuses that we have not been able to put technology so readily available into planes, especially when most of the tech required for such a system is already in use for giving passengers in flight wifi....
Link towards satellites is reliable if directional antenna is "pointing up" so it's not good if airplane is falling
In many (most?) cases the most important data retrieved from a black box is not in recording the catastrophic final moments where a satellite connection would likely be broken. It is in finding clues as to to the cause of the accident by analysing voice/data in the run up to it.
I have the feeling, that in aviation always something has to happen, until something happens!
Can you talk about instances, where engineers thougt through systems, procedures and improved them before something happened?
Your great videos on the Mentour Pilot channel sometimes give me that negative connotation, that always something has to happen until things get improved!
Well, maybe because it's so expensive to introduce an innovation into industry? And certify that.
We need Mentour Engineer channel! 😂
Thank you for the upload. Have a great week. Suggest could avoid the transition effects, they are disturbing a bothersome?
That "burst transmission" idea is what I was typing up while watching this video, because compression and burst transmission cut a lot of the overhead of "streaming" down. And no, replacement of black boxes sounds like a bad idea. Augmenting them with extra capabilities like burst transmission sounds great. If we can get it back to 5 minutes of data in that burst, that probably takes care of most of the technical incidents I've seen on this channel, because they seem to occur very quickly.
I came bk to this video after the Alaska airline issue.
Another argument for.
Arguing against redundancy seems a little silly to me.
Especially since it's with something this relatively simple.
Split the data stream, including voice recorder and send one copy off the vehicle in an asynchronous manner.
More data=more better.
It doesn't have to be perfect, and the cost is not going to be that high as long as you're ok with a best effort channel rather than a dedicated guaranteed data channel.
20GB over a 4 hour flight is "only" 12Mb/s.
Way less than what the in flight entertainment systems use.
Besides. Data can be compressed.
I would be very surprised if you couldn't get the data stream down to under 1-2 Mb/s.
Just my two cents
I agree. It should not be a question of one or the other. Do both. If data stream was cut, there is still the chance that the black box could be recovered. Starlink will definitely make it more likely to be doable.
Turning 12mb into 1-2mb isn't really possible unless the data is very repetitive. Plus, compression takes CPU power and time, which means that if it was compressed it would take longer to be ready to be sent off to servers.
@@YHDiamond And luckily, audio is very compressible. MP3, for example, has an 11:1 ratio. More application-specific methods could be used to compress things like the aircraft's position, speed, and other things. As for CPU power and time, by watching this YT video, you forced your computer to perform some pretty extreme compression and decompression, in real time. Even on an a VM on an old chromebook, with a Celeron CPU and other stuff in the background, I was able to compress stuff at bandwidths way higher than 12 Mbps, so it would easily be possible with modest computing power.
@@RyanTosh I'm not sure if they would use a lossful compression algo
@@YHDiamond so i just did some back of napkin calculations.
88 channels, with a sample rate of 1000Hz, half with 8 bit precision and half with 16 bit precision is just a hair over 1Mb/s. Add 4 channels of voice data at 64K and we're still way under 2Mb/s.
And yes. I know there are a lot of assumptions in my calculations.
But i don't think that they're that far off.
And I'm pretty sure that most data is repetitive enough to be compressed.
You don't really need 1000 discrete samples of engine speed. I'm fairly certain that a compressed second would be more than adequate.
But again. That is just my assumption and i realize that
My guess is with thousands of planes in the sky at any given moment, the amount of data to deal with would just be phenomenal..... (commented before watching the video) PS I JUST booked my flight YUL to YVR for July 1st with my cat WOOHOO and i bought a vintage Air Canada tail fin pin off Ebay that I'm gonna wear then ❤️
our modern world can handle it though
This is not as crazy as the number of people who stream videos in the plane while it’s flying
And you only need to store it for just a max of a month
@@c4fusion1 it's because the aircraft is already fitted with different satcom antenna. I guess you already know which one is it.
And as a guy from a broadcast background, most videos you watch right know are encoded and serve you as compressed-ready-to-watch format (that's why the bandwidth cost are so small, and there's no interference in terms of watching experience), while flight data are mostly raw data from the sensors that even the most powerful smartphone couldn't decode it in seconds. And most of the time, on poor connections, video streaming would just buffer before it receives valid data. Flight data streaming does not work that way. If you experience packet loss/high ping due to insufficient wireless transfer rate then the received flight data becomes unreliable. Also adding tools like this means the operating cost of the aircraft increases.
Why does it have to be a different antenna if we are just using this as backup to ensure that we have data? We can use the internet to transfer the data, and before you say it's too sensitive, why would it be? The world's financial data is run through the internet.
As for the bandwidth, how is it not enough? There is only 5 voice streams and 38 parameters. Let's say the 38 parameters need to be recorded at millisecond precision, it's still only 1.216 mbs (assuming the parameter fits in 4 byte, which if it's a number it will) or about tenth of the bandwidth for 1080p youtube video. As for compressing voice data in real time, a cellphone from the 2000s can do it digitally much less a modern cellphone (incidentally, a modern cellphone would probably have more processing power than the flight computers in most airplanes).
As for the video streaming comparison, I wasn't making a point of what equipment you need (though you can live stream using h.254 via a smartphone these days) as much as how much bandwidth you need. Since satellite internet unlike cable internet is symmetrical, the down speed would be equal to the up speed. The nature of the internet
As for the packet loss issue, there are many ways to ensure high availability of your connection such as packet retrying that is built into most streaming applications such as the youtube video you are watch. Even better yet, I know of a sewage treatment plant that has a satellite plant and pumping station that is controlled and recorded remotely. This being a backup would not need as high of a reliability and if it did, the tech is already available and is at least 10 years old.
Along with voice recorders, I feel like cockpit, cabin and external video recording should be a worthwhile next step to include in planes. They'd help investigate things like bird strikes, engine fires, pilot behaviour during an emergency, or even passengers being dangerous. It wouldn't just be useful for visually investigating a lot of issues before they escalate into crashes.
We do not need to stream all of the Black Box data to the cloud. We need only to send a small packet of the GPS location, speed, and direction every 30 seconds for aircraft over oceans or remote areas. If there is a crash that will help us find the Black Box more quickly and maybe survivors.
Teah. I support backups of data. I would have reccomended a smaller light version of a black box that maybe records the last 30 min of flight data and 10 min of audio, rather than the standard black box that records in 20 hours and 2 hours respectively.
This lighter black box should be stored in a separate place of the plane with less protection from being damaged because, let's face it; making theese things really robust to the point where they can withstand direct impact with the ground is really expenside. A plane normally isn't 100% pulverized in a crash. There will be some pieces of debris that end up with minor to no damage. This light version is just there in case the regular black box had bad fortune and got destroyed, that it might hopefully could have the good fortune of being in a place where it does survive.
There's still the posibility that the plane might land in the ocean and nothing be recovered. So a transfer of data through radio or simething is a good idea. But such data transfer is costly in large amounts. So, yes. Only small amounts of data bursts in case of the plane exceeding certain pre-determined parameters would be more than enough help for investigators to try and discover the cause of the problems. Anything is better than nothing.
I think uploading to the cloud is a great idea. I mean, it's just out the window in IMC.
I think it kind of goes without saying that a streaming flight data system would be supplemental, and not replace the black boxes. With systems like Starlink, the cost of net data transfer is going to continue to drop until it gets trivial. Especially in the most extreme examples (such as the recent nosedive of China Airlines 5735), a satellite link is probably going to lose contact due to the attitude of the aircraft.
as satellite internet decreases in price, the "cloud black box" will become the industry standard. Just a matter of time and economics. However, during the MH370 investigation, I remember there being a segment where they mentioned the engine manufacturers such as GE and in MH370 case Rolls Royce receive periodic updates as to engine performance from their control centre in the UK via Data link. infrastructure is in place, just need a "Henry Ford" to make it affordable for all.
and in any color the customer wants, as long as it's black.
That data link is quite different from anything that can support full flight and voice data livestreaming. Remember that for this to work you need 100% coverage, so having 'enough' satellites for full coverage is not enough, you'd need full overlap in case a satellite doesn't work or is overloaded etc.
Even Starlink would not be capable of this currently, perhaps in another 10 years at the soonest.
The amount of data, lets take the 20GB per flight Petter mentions as an average means that if we only keep a rolling 24 hour period there's plenty datacenters that can store that without issues, but getting the data from the aircraft to that datacenter, mid-flight is not possible right now, not 100% of the time anyway.
Then the biggest issue, you mention 'just a matter of time and economics' but which economics? Where is this ever going to save money let alone make money? It's a hole in the ground which will never earn money. It perhaps could save part of the cost of searching for a lost aircraft as we would just know where it is but the entire cost of the system would be hundreds times more than those potential savings... This only saves money if we have a missing crashed aircraft every day or so.
@@someguy4915 once upon a time, we had dial up modems, having a cloud account like google drive or Dropbox seemed like an unnecessary construct. Downloading or uploading 100mb seemed like an eternity.
At that time - if I told you that homes and apartment blocks would be connected with fibre optic cables, you would rightly call me insane!
Yet, here we are.
@@someguy4915 with regard to the economics - off the top of my head, firms like Lockheed Martin or Raytheon or even Boeing need to start mining data for their autonomous flight programs.
Whilst drones are just in the military spheres right now, in time it will filter down to commercial airliners as well.
@@ahmedmotala7282 Don't expect pilot-less airliners to become a thing within the next 30 or so years, public opinion would have to be changed first, then legislation and then they also have to build it.
Besides that, I doubt that some blackbox data will really help in that field without the full image, the blackbox reporting a rough landing is useless without weather information at the time and without a report of the pilots explaining that there was a microburst that the weather-radar can't see etc.
But even if this could be useful for data gathering, that's no reason to deploy it on all airlines, just on a select few where they can get the side-information from pilots etc.
And if any single one of those aircraft crashes the first question the public will ask is if that new 'drone passenger plane' program had caused it, destroying the chances of such aircraft ever being accepted by passengers.
TL;DW: because, especially since crashes rarely happen, it's too much data (more than about 20gb worth per airplane per flight) that needs to be sent all the time across all flights. The last 20 hours of flight data and the last 2 hours of high quality voice communication need to be recorded in order to determine what happened in the event of a crash.
I’m a software engineer, every single request made to the web server is almost always logged, so one person opening one webpage could have 20+ long lines of logs, and each interaction with the web server is logged in one form or the other, these data are preserved for much longer than 20 hours.
I really like the short burst idea. With the position data combined with attitude, I'd imagine the crash site could be nailed down very accurately. What I don't get tho is why the cvr still only stores 2 hours. I get that it's probably not a grocery store level sd card in there, but still. Is this special storage that expensive?
Id imagine its probably something to do with making it durable. Can only fit so much tape with all the armor around it. This is just my best guess btw.
No it isn't. Even high quality flash used for missiles is dirt cheap compared to even the metal box that encloses the electronics. We actually fit much bigger flash than we need because the smaller ones aren't made much now. The night vision goggles I worked on had 8GB of flash. We could have stored a million seconds of phone quality uncompressed audio, or 200,000 seconds of high quality stereo. That's over 2,500 and over 500 hours respectively.
When film or tape was used it was a technical problem. When it got digital in solid state chips, it was no longer a technical problem, but a political problem.
The pilots in USA has insisted for years that it should be no longer than 2 hours.
There was an accident with a cargo ship several years ago which steamed into a hurricane, by law they need a couple hours of data and voice recording on the bridge, but by sheer luck, a much larger SD card was used, so something like 15 hours of voice recording was available, and was very helpful to understand the decisions which led to the sinking. Of course, ships take much longer to get anywhere than planes, but records of an entire flight could be very helpful indeed.
@@a4d9 It was wire before that, because wire survives crashes and fires. All FDRs should be switched to solid state now, tape wears out too fast, and breaks easily. Wire was tough but couldn't record much.
My thoughts: tail is wagging the dog insofar as data recording goes. The solid, on-board works and takes decent amount of engineering to survive a physical shock, But, it is relatively obsolete. It should continue but should be buttressed by some modern technology. Even if the technology isn't 100%, This kind of data is retrospect and has nothing to do with flight control...it is forensic only, The more forensic data you have the more likely you'll arrive at a proper conclusion. Even if the 'real' black boxes drown, you'll have some data. Oh yea, FAA or whoever would say it needs to be 100% reliable and without gaps. But does it really need to be?
Given that the reputations of the living and the dead will be on the line they'd be right to insist that it be reliable and without gaps. We can aspire to Just Culture all we want, but unfortunately people in pain demand to know who to blame (even if they end up blaming the wrong person).
@@DrewNorthup
0
To me there seems to be a privacy reason, too. Imagine all flights are readily available this might encourage some airline managers (think Ryanair) might want to improve their performance by using that FDR data in an automated fashion from each trip. This could make it even more uncomfortable for pilots and generate useless stress as they could link everything to the actions of everyone.
That data is already available to the public. The only way one has privacy while exercising the privileges of a pilot certificate is maintaining visual flight rules, or in some areas flying an aircraft without an engine-driven electrical system under visual flight rules. How do you think you can follow aircraft by registration number on various websites?
@@r2db That is ADS-B data but as far as I understand it does not have as many parameters as raw FDR data. And of course you can see some performance data like F/L, speed and direction but (correct me if I am wrong) no real direct steering inputs and other data that are not sent by ADS-B. Like rudder/flap/thrust settings It is also only updated every half second which sounds like it does strip out lots of data. I think it's ok if anyone can follow a plane but if someone would look at every keyboard press I made in my job I would feel quite a bit more intimidated.
@@gentuxable My point is that your privacy concern is already a lost cause and has been for some time. It did not exist for aircraft on IFR flight plans pre-ADS-B and it doesn't exist now for any ADS-B capable aircraft. Some engines in many different industries have telemetry data that already communicates regularly if not continuously with the manufacturer (as an example, the Cummins X15 where it has the capabilities for over-the-air parameter updates). I have never flown an Airbus but it seems that pilots of their aircraft have been aware of telemetry data it sends for some time, and farm tractors such as some John Deere models also have OTA capabilities that allows the company to disable them if they are stolen or involved in some present wartime action. In healthcare, every mouse click and every keystroke is logged with a time stamp. As a pilot, if data is sent over the aircraft bus to the FDR then there is nothing preventing a separate piece of equipment from logging or transmitting such data for corporate purposes.
If you do not want someone looking over every keypress you make then in today's environment the best suggestion would be not to take a job involving anything with a computer in any form.
@@r2db Ok I get what you say but to me it seems like it still goes within a reasonable range with ADS-B and clearly there are some companies that go far beyond what we outsiders know of. We're getting into dangerous waters in almost every aspect (which I'm sure would also lead to some ruling if court at least in Europe knew and that is often an issue already to bring it out). Having the FAA (or EASA for that matter) mandate all 88 FDR uploaded to a server seems to go further than everything done nowadays I'd still think and it would send undesirable signals to the industry.
BTW if my company records every keyboard stroke I'll sue them, I know my rights and that is a clear violation.
Well it's already mentioned in mentour pilot channel regarding pilot privacy. And most pilots agree if their airlines use flight monitoring system like ACARS. Still wishing for privacy? You know that lavatories can be opened from the outside right?
The QARs you mentioned can do this. They send recorder data to an airline HQ via 4G cell service. If a pilot lands hard they know, an HQ can also use the data to save fuel (make better flight plans with different speeds).
Also recorders still work during an electrical failure. The battery supplies essential power. However it may not broadcast anywhere since that works off main power.
Thank you, as always, for the enlightened analysis. Fly safe, Capt. Petter!
Not really impressed by the data transmit volume argument. It could be cut into tenths and give a better clue than nothing at all.. Not crying for the airline's expense. They sell the same game to customers(internet). Really now. Data recorders are old and very good but, they're old. So am I. The data X-fer is not much given the tech now.
Well whether the cost is high or not the cost will be passed down to the customer and as it's so rare to need the black box and not have it, I'd rather not pay for the cost of sending the data I'm not going to need rather than paying for it on every flight just to have a less than 1% chance that's going to be the flight. I'm in the US and flights are already expensive enough here.
I think a better case could be for private aircraft since they have way less data to send, but the data we could get out of that could be just as valuable.
I wished everyone studied broadcasting, aviation, but ok I'd pass your ranting for this.
Didn’t Qantas headquarters have alarms go off and knew that flight 32 had trouble? Whatever data it was, something was being monitored in real time.
They only care about fuel and location not other stuff
Any redundancy on power, duplicate black box? I would add data streaming when pilots hit the panic button, not just a short burst!
Pilots have better things to do than hit a panic button.
@@cageordie , it's a figure of speech. Nevertheless, would be nice if the streamed data can be share with ground control in real time.
@@test40323 No reason it can't be, other than nobody having really asked for it yet. All this about ADSB is just a sideshow, there are plenty of communications satellites, they just need to work out what they want. But anything requiring pilot intervention is a non-starter. There was a great interview with Chuck Yaeger about being a test pilot and the bit I remember was him describing seeing a guy in a crashing plane "The guy is running all over the cockpit writing notes and talking on the radio. That's all BS. You are just trying to save your life." That's the way it is.
@@cageordie , good point. But some emergencies have more time and if the ground crews can help figure out how to resolve it based on the live data ... the work load would be eased by the extended team.
@@test40323 But then those aren't the sort of emergencies where you are digging the black box out of the wreckage to find out how it came to be there.
I'm not a pilot but I learn a great deal abt all aviation related topics from you
TY
In two words: Data Analytics
A plausible reason why air lines would adopt full flight data upload to the cloud is regulation. If it were regulated this way then the airlines would not have a choice. In this case the cost of the data uploads would be reduced by economies of scale. If I had my way, safety orientated systems would be charged at cost.
The question why would you do this and you can't just say Data Analytics?
The main reason I would suggest is that with each flight storing it's data in the cloud, historic flight data could be studied.
In that way, in the event of the loss of an air craft whether the Black Box is recovered or not, the flight data is able to be studied to try to determine where a fault first showed signs of developing before the incident. Or alternatively, when pilots begin reporting a new fault, then the current data may be studied and correlated between aircraft to detect the fault and trigger pre-emptive actions on air craft where the fault hasn't yet been reported.
Other studies of historic flight data are available.
P.s. This or something similar may have been mentioned in an earlier comment but I haven't read all 1000+ comments already posted.
I am an electronics engineer and a pilot, and i completely agree with your analysis! We will need a new satellite network just to track all data from all planes out in the sky all over the world at the same time...
Yeah, especially voice data needs really much bandwidth. How much data (and which data?) is transmitted via ADS-B? I guess with increasing bandwidth availability more parameters could be transmitted this way.
@Alfred Weber I doubt you already learn anything about networking and broadcasting.
Ever heard of the Kessler Syndrome?
@@annando ADSB data is transmitted only when the transponder is interrogated by ground radar or other transponder. Usually between 1 per few seconds up to few times per second. It requires low bandwidth, since transmitting very few parameters , at very low rate.
@Oferb553 BTW: I just saw the latest video by Scott Manley and coincidentally he spoke about ADS-B. He mentioned that Canada now requires a space based ADS-B for planes in their air space.
As a complete casual and non-expert (even "amateur" might be pushing it...), what I find disconcerting about aviation isn't the black box issues. As it was said, they are quite reliable in most cases.
But what is apparent to me in a lot of the videos here and elsewhere, is how unreliable the day-to-day voice communications are. From relying on English globally (or any one language no matter what it is), to either controllers or pilots missing an instruction or getting it wrong, or having interference garble a transmission, it feels like there's got to be a better way to transmit and acknowledge instructions more reliably.
Very true! At the very least, there should be voice recognition software in every cockpit so that instructions are not only heard through the headphones, but also appear as text on a screen. Ask any pilot whose first language isn't English, and who has struggled to comprehend instructions rattled off by a guy in a control tower at 1,000 words per minute (or thereabouts.)
@@nicolasuribestanko A text message is a good idea, but voice recognition is not. Voice recognition are not reliable at all and could cause a lot of confusion if anyone was relying on it and it transcribed wrong, which it definitely will, which will cause serious issues for safety when it comes to receiving instruction from ATC.
What is far worse than using one standard language, with a standard set of phrases? Not having standards. I earned my pilot certificate in an era of "position and hold" and still dislike "line up and wait" but it is what it is and I can't change it. If I read back the instruction as "position and hold" I am not properly communicating. Why is voice communication still the gold standard? Because when one is flying an aircraft, particularly during an emergency, one does not have time to type on a keyboard. Both hands and both feet are frequently in use during an emergency, and voice communication takes one finger with an occasional twist of a knob or button push to change frequencies. There is also no space in the panel of an average general aviation aircraft, and definitely not in the flight deck of a heavy transport aircraft, for a bunch of additional buttons for communication.
What I always wondered was, why isn't there a camera on the flight deck?
Privacy
@@AgentOffice 🙄🙄🙄. What privacy? You're at work!! Smh
@@christophermercado5466 police were anti body cam too
@@AgentOffice bro stop, please stop!! Why do you think they didn't want to be on camera. Just stop
Cloud storage could be a backup, not a replacement. The backup stream could focus on just the changes or the most critical parts and with the latest compression algorithm this is definitely possible to do.
They should have a cam in the cockpit and a tracking device on the black box. Everything can always be improved.
We don't need to have real-time transmission of data and voice record to replace the black boxes all the time. I agree that operational and cost consideration must be balanced with safety needs. At the end of the day, we cannot and do not aim at eliminating ALL risks. We minimize it with minimal cost and operational difficulties. The "operational data monitor program" you talked about can be further enhanced. We can also make it to be triggered by alert events (e.g. when a master caution or GPWS warning or stall warning is issued, etc., or even when the voice recording detects the pilots said "PAN PAN" or "MAYDAY" in the transmission to the ground, etc.). We can also make that "short burst of data transmission" to contain more data (e.g. contain more parameters and even voice recording, or with data in the last 30 min instead of 2 min). Lastly, we can also make the transmission real-time since the alert happens, so the "cloud" will receive continuous data since the alert happens (instead of just a short burst of history for a few min before the alert) for the next 30 min. Things can be improved this way further, rather than a complete replacement. We actually rarely replace all technologies completely. For example, we still have telegram and fax available, and bank branches are still everywhere despite online and mobile banking, etc.
I'm not sure, to answer the question. If I had to guess, it would be $....
In any case, over the Indian ocean there is no radar observation. But....Over the Atlantic? Despite the fact that all the airplanes use CPDLC to communicate their position in space to Gander/Shanwick, there ~~IS~~ radar coverage over the north Atlantic! This is specialized air defense radar, and we don't want the filth in Russia to know it's capabilities, so your plane is officially not in radar contact. Even if it actually is...
The size of data needed to be sent like you said 20 gigabytes for example could be reduced massively by using some new file compression tech. Probably down to single gigabytes and then when it would be viewed it could be decompressed to have the original quality and size.
I think the reason streaming black box data hasn't been implemented is just because companies are greedy to make money and want to avoid improving anything that isn't absolutely necessary by the law.
20GB is not a lot for LEO satellite constellations like starlink. That’s probably on the order of a couple dollars. Plus each flight can have hundreds of tickets to pay for that
The housings for commercial & military black boxes used to be made by Valley Aluminum Works in Phoenix, Arizona. They were made from hot forged aerospace grade aluminum. From VAW they went to Harvey's Deburring, then to a machine shop, back to Harvey's for more deburring, & degreasing and then on to an enamel powder coat paint shop for a hot baked on orange enamel coating.