Thank you, Prof Emil and Prof Eric for the amazing podcast. I would like to add a few reasons which I think were also instrumental for migration toward more TDD deployment as compared to FDD and would love to get your insight on this. 1. Flexible UL/DL Configuration. When 4G was launched, it was expected that the DL Data rate is a lot higher than UL, but with time it appeared that UL Data rates were also increasing at a faster rate compared to DL (video conferencing, social media, gaming, etc). This means it would be more efficient to rely on TDD than FDD but 4G suffered from fixed UL/DL Configurations (meaning they were static for cell). 5G got rid of this limitation by introducing dynamic UL/DL configuration thereby giving full freedom to dynamically change how much rate to support in DL vs UL. 2. Cost for Front End RF components reduce by half FDD requires a separate antenna, filter, oscillator, amplifier, etc for supporting simultaneous DL and UL streams; along with more processing power on the baseband signal. TDD doesn't require separate components for DL and UL but a need for an additional component of a switch to toggle the direction. Cost-wise, TDD ended up becoming more cost-efficient as the RF components are one of the costliest components (especially for the mmwave spectrum).
Thank you for your input! I think what you are saying is correct. Here are some brief comments: 1. We certainly have more flexibility in TDD since we can select the UL/DL configuration by software in TDD, while it is determined by the spectrum assets in FDD. However, dynamic UL/DL configuration doesn’t seem to be on the roadmap. To avoid “strange” interference variations over time, the licensees of neighboring bands are supposed to be synchronized and use the same UL/DL configuration. In Sweden, these the DL/UL ratios are specified in the licenses so they cannot be changed. The ratio is selected for DL heavy traffic so it isn’t good for UL heavy traffic (e.g., a network of surveillance cameras). 2. I’m not familiar with the cost aspects, but it makes sense that TDD requires fewer components thanks to the half-duplex mode.
@@WirelessFuture Thank you for your prompt response. I would find it very strange that the licensees would restrict the UL/DL configuration defeating the purpose of advancements made with 5G. Below is my observation based on prior work experiences... - The 3GPP spec defines the use of guardbands, which I assumed was to minimize intercell interference. There would definitely be an issue for Contiguous Carrier Aggregation (CA) cases, but you don't expect different network service providers to configure CA in this fashion unless there is a very tight business and technical collaboration. Within a single network service provided, Contiguous CA is quite possible and synchronization is needed to avoid interference as there is no guard band. The spec also covers in a lot of detail the ACLR (Adjacent channel leakage ratio) requirements. It is fair to recognize that early 4G deployment might be risky as OFDM was used for the first time but 5G uses the same physical layer waveform, so I would expect maturity in Hardware and RF components to avoid bad leakages and intercell interferences. Reference : 3GPP TS 38.101-Release 17 1 V17.6.0 (2022-06) , Section : 5.3.3 Minimum guardband and transmission bandwidth configuration - 5G (as opposed to 4G) introduced the concept of flexible slots where the slot can be made either UL or DL depending on the type of grant (DL grant or UL grant in PDCCH). Of course, this is keeping in mind that 1 or more slots might be required for DL/UL or UL/DL transition. Since there is RB level allocation, it is possible for networks to keep some spacing at the edges of the band to avoid or reduce intercell interference if required. This freedom didn't exist in 4G hence it makes sense to specify and restrict the DL/UL ratios in the licenses. Reference: 38.213 v15.7 -Table 11.1.1-1: Slot formats for normal cyclic prefix Secondary Reference: www.sharetechnote.com/html/5G/5G_FrameStructure.html
14:54 In China you can do almost anything regarding spectrum. Thank you Prof. Emil Björnson for promoting massive MIMO with TDD. In addition, I thank you for making channel reciprocity very easy to understand. Great discussion. I like these kind of podcas
What is commented by prof. Björnson is rigth, on USA is deploying Massive MIMO on LTE and NR 2.5GHz TDD, for band 600MHz also is using LTE/NR FDD but only with a configuration 4x2 or 4x4 MIMO , T-Mobile is using 2 vendors, Ericsson (most on west coast) and Nokia (east coast). Good podcast, thank you.
Thank you for this great discussion. I started a short time ago my PhD thesis on analyzing and improving signal processing Techniques in Cell - Free Massive MIMO system, and I suggested this as a topic for discussion .. With all thanks
"There is a debate about TDD/FDD" 16:33 "Of course: as we know TDD is fundamentally better" Nice to see you playing it safe and not naming a favorite :-D
Thanks for this initiative. It is such a good platform to know about the latest in massive MIMO and future wireless systems. I would appreciate if there can be a podcast on 'Channel State Information acquisition in TDD massive MIMO systems using deep learning'. Looking forward to more of these podcasts.
Probably one of the best podcast in the Wireless Technology subject,,, really enjoy the podcast. I hope to finish watching all the episodes and maybe one day I will study at KTH.
I was smiling when I read our the author of our course assigned book for MIMO. Very interresting to have come across this channel. As you are very specialised in RF, I have a short question. What is your opinion on the long term job perspectives for people wanting to work and design RF components/systems. Are we reaching hard limits soon on how far we can take RF technologies and how much new advances are actually needed.
I think there is much more to be done. Communication theorists are much faster with coming up with new concepts than it takes to develop efficient implementations. Millimeter wave implementations are rather inefficient, and new challenges arise as we continue up in frequency or when we try to integrate energy harvesting into the network.
Dear Emil, Thanks for your informative post. If I correctly understood you are saying that 16T and 16R means the number of RF chains. The question is Is it essential to have an equal RF chain and baseband chain?
I'm not sure what you mean by "baseband chain". After/before sampling, the computations happen in a baseband computer. However, if you transmit 4 signals using 16T/16R, then most of the baseband processing can be done as 4 separate computations, and then we use beamforming vectors to transform it to/from 16-dimensional signals.
There are products which marketed as FDD MM which are even available for 4G let alone 5G. Could you please cover what’s actually behind FDD MM and how is it different to TDD MM
Thank you, very much!! One question please, few years ago we did trial with LTE TDD 2600 B38 with 64T64R with reciprocity based BF and we found that with 100 users in cells were limited by SRS capacity to 32 users. do you know will it be the case in NR also?
Thank you! I don’t know the exact details of the standard, but I’ve heard about similar limitations in NR. If it becomes a practical bottleneck, I suppose that 3GPP will address it in future releases.
![Ep 2. Myths About Massive MIMO [Wireless Future Podcast]](http://i.ytimg.com/vi/Aut3YMhHAEw/mqdefault.jpg)
![Ep 2. Myths About Massive MIMO [Wireless Future Podcast]](/img/tr.png)
![Ep 22. Being Near or Far in Wireless [Wireless Future Podcast]](http://i.ytimg.com/vi/7aVCtTgyMx4/mqdefault.jpg)
Dr.Bjornson is the best professor in Massive MIMO that i have ever seen. Thank YOU veryyyyyyyyyyyyy much.
Thank you, Prof Emil and Prof Eric for the amazing podcast. I would like to add a few reasons which I think were also instrumental for migration toward more TDD deployment as compared to FDD and would love to get your insight on this.
1. Flexible UL/DL Configuration.
When 4G was launched, it was expected that the DL Data rate is a lot higher than UL, but with time it appeared that UL Data rates were also increasing at a faster rate compared to DL (video conferencing, social media, gaming, etc). This means it would be more efficient to rely on TDD than FDD but 4G suffered from fixed UL/DL Configurations (meaning they were static for cell). 5G got rid of this limitation by introducing dynamic UL/DL configuration thereby giving full freedom to dynamically change how much rate to support in DL vs UL.
2. Cost for Front End RF components reduce by half
FDD requires a separate antenna, filter, oscillator, amplifier, etc for supporting simultaneous DL and UL streams; along with more processing power on the baseband signal. TDD doesn't require separate components for DL and UL but a need for an additional component of a switch to toggle the direction. Cost-wise, TDD ended up becoming more cost-efficient as the RF components are one of the costliest components (especially for the mmwave spectrum).
Thank you for your input! I think what you are saying is correct. Here are some brief comments:
1. We certainly have more flexibility in TDD since we can select the UL/DL configuration by software in TDD, while it is determined by the spectrum assets in FDD. However, dynamic UL/DL configuration doesn’t seem to be on the roadmap. To avoid “strange” interference variations over time, the licensees of neighboring bands are supposed to be synchronized and use the same UL/DL configuration. In Sweden, these the DL/UL ratios are specified in the licenses so they cannot be changed. The ratio is selected for DL heavy traffic so it isn’t good for UL heavy traffic (e.g., a network of surveillance cameras).
2. I’m not familiar with the cost aspects, but it makes sense that TDD requires fewer components thanks to the half-duplex mode.
@@WirelessFuture Thank you for your prompt response. I would find it very strange that the licensees would restrict the UL/DL configuration defeating the purpose of advancements made with 5G. Below is my observation based on prior work experiences...
- The 3GPP spec defines the use of guardbands, which I assumed was to minimize intercell interference.
There would definitely be an issue for Contiguous Carrier Aggregation (CA) cases, but you don't expect different network service providers to configure CA in this fashion unless there is a very tight business and technical collaboration. Within a single network service provided, Contiguous CA is quite possible and synchronization is needed to avoid interference as there is no guard band. The spec also covers in a lot of detail the ACLR (Adjacent channel leakage ratio) requirements. It is fair to recognize that early 4G deployment might be risky as OFDM was used for the first time but 5G uses the same physical layer waveform, so I would expect maturity in Hardware and RF components to avoid bad leakages and intercell interferences.
Reference : 3GPP TS 38.101-Release 17 1 V17.6.0 (2022-06) , Section : 5.3.3 Minimum guardband and transmission bandwidth configuration
- 5G (as opposed to 4G) introduced the concept of flexible slots where the slot can be made either UL or DL depending on the type of grant (DL grant or UL grant in PDCCH). Of course, this is keeping in mind that 1 or more slots might be required for DL/UL or UL/DL transition. Since there is RB level allocation, it is possible for networks to keep some spacing at the edges of the band to avoid or reduce intercell interference if required. This freedom didn't exist in 4G hence it makes sense to specify and restrict the DL/UL ratios in the licenses.
Reference: 38.213 v15.7 -Table 11.1.1-1: Slot formats for normal cyclic prefix
Secondary Reference: www.sharetechnote.com/html/5G/5G_FrameStructure.html
14:54 In China you can do almost anything regarding spectrum. Thank you Prof. Emil Björnson for promoting massive MIMO with TDD. In addition, I thank you for making channel reciprocity very easy to understand. Great discussion. I like these kind of podcas
What is commented by prof. Björnson is rigth, on USA is deploying Massive MIMO on LTE and NR 2.5GHz TDD, for band 600MHz also is using LTE/NR FDD but only with a configuration 4x2 or 4x4 MIMO , T-Mobile is using 2 vendors, Ericsson (most on west coast) and Nokia (east coast). Good podcast, thank you.
It's extremely helpful to all who are seeking information about Massive MIMO.
Thank you for this great discussion. I started a short time ago my PhD thesis on analyzing and improving signal processing Techniques in Cell - Free Massive MIMO system, and I suggested this as a topic for discussion .. With all thanks
"There is a debate about TDD/FDD" 16:33 "Of course: as we know TDD is fundamentally better"
Nice to see you playing it safe and not naming a favorite :-D
Thanks for this initiative. It is such a good platform to know about the latest in massive MIMO and future wireless systems. I would appreciate if there can be a podcast on 'Channel State Information acquisition in TDD massive MIMO systems using deep learning'. Looking forward to more of these podcasts.
Probably one of the best podcast in the Wireless Technology subject,,, really enjoy the podcast. I hope to finish watching all the episodes and maybe one day I will study at KTH.
Thanks a lot for this presentation
Such an insightful podcast.
Extremely interesting podcast! Please keep up, it's what we want here!
Great discussion. Very appreciated
I was smiling when I read our the author of our course assigned book for MIMO. Very interresting to have come across this channel. As you are very specialised in RF, I have a short question. What is your opinion on the long term job perspectives for people wanting to work and design RF components/systems. Are we reaching hard limits soon on how far we can take RF technologies and how much new advances are actually needed.
I think there is much more to be done. Communication theorists are much faster with coming up with new concepts than it takes to develop efficient implementations. Millimeter wave implementations are rather inefficient, and new challenges arise as we continue up in frequency or when we try to integrate energy harvesting into the network.
Thanks to the professors such a information.
Many thanks for the helpful discussion!
Dear Emil, Thanks for your informative post. If I correctly understood you are saying that 16T and 16R means the number of RF chains. The question is Is it essential to have an equal RF chain and baseband chain?
I'm not sure what you mean by "baseband chain". After/before sampling, the computations happen in a baseband computer. However, if you transmit 4 signals using 16T/16R, then most of the baseband processing can be done as 4 separate computations, and then we use beamforming vectors to transform it to/from 16-dimensional signals.
This is great
Very interesting! Thanks again!
Very informative
Thank you Sir😊
it is nice initiate...i like and enjoyed it...keep it up...
There are products which marketed as FDD MM which are even available for 4G let alone 5G. Could you please cover what’s actually behind FDD MM and how is it different to TDD MM
TDD decoding is simple than FDD , because TDD has a inverse property so easily zero forcing algorithm could be applied
Great personal
Thank you, very much!! One question please, few years ago we did trial with LTE TDD 2600 B38 with 64T64R with reciprocity based BF and we found that with 100 users in cells were limited by SRS capacity to 32 users. do you know will it be the case in NR also?
Thank you! I don’t know the exact details of the standard, but I’ve heard about similar limitations in NR. If it becomes a practical bottleneck, I suppose that 3GPP will address it in future releases.
pls do talk about how vendors are tackling mm-wave
We now have a new episode about mm-wave: ruclips.net/video/dwtMtRE8Y58/видео.html
I have seen all episodes waiting for new ones after 7 th.
Thank you for watching! We release a new episode every other Wednesday.