MIMO 2x2 to 4x4 Change

[umutaltun13]

Hey everyone,

I have question about MIMO. What are the disvantages you see in your network of changing MIMO 2x2 to 4x4 ?


Thank you in advance.

[remoz_A]

mainly will be cost and installation

[longtermevolution]

If you have fixed RRU power which is now shared between 4 ports rather than 2 ports your coverage will shrink. if still having same power per port you will not face any negative impact.

Spatial layer cause mutual interference to each other so probably you might face:
SINR degradation, lower MCS and hence degraded capacity per layer compared to 2X2

[firstmaxim]

Less than 1percent of UEs in a cell would actually use MIMO 4x4. 50 percent of UEs. Only the UEs very close to the base station can be served with 4x4 MIMO because of the high S/N requirements.

[subway]

Less than 1percent of UEs in a cell would actually use MIMO 4x4. 50 percent of UEs. Only the UEs very close to the base station can be served with 4x4 MIMO because of the high S/N requirements.

To add to that, the fixed CRS overhead of your network will rise form 11-12% (2x2 MIMO) to almost 25%.

In my opinion, 4x4 MIMO only worth it where:

1. The amount of 4x4 capable UEs is high (at least 40-50%).
2. The sites are rather close to each other (dense urban).
3. And you have good reflective built environment.

In any other case, you will spend a lot of money, introduce a fixed speed penalty for most customers (CRS overhead), and will see very little gains.

To give you an example, our network is a rural WTTx one, everything is 4x4 capable yet it does not worth to turn it on.

If you can, think of beamforming, high order modulation (256QAM DL, 64QAM UL) and such instead of 4x4 MIMO. For example if you can do TM9 you can also exploit 4 spatial layers (compatible UEs required and a minimum of 8T8R antennas), but without the added fixed CRS overhead of the traditional 4x4 MIMO.

[felixj.alvarez]

To add to that, the fixed CRS overhead of your network will rise form 11-12% (2x2 MIMO) to almost 25%.

In my opinion, 4x4 MIMO only worth it where:

1. The amount of 4x4 capable UEs is high (at least 40-50%).
2. The sites are rather close to each other (dense urban).
3. And you have good reflective built environment.

In any other case, you will spend a lot of money, introduce a fixed speed penalty for most customers (CRS overhead), and will see very little gains.

To give you an example, our network is a rural WTTx one, everything is 4x4 capable yet it does not worth to turn it on.

If you can, think of beamforming, high order modulation (256QAM DL, 64QAM UL) and such instead of 4x4 MIMO. For example if you can do TM9 you can also exploit 4 spatial layers (compatible UEs required and a minimum of 8T8R antennas), but without the added fixed CRS overhead of the traditional 4x4 MIMO.

Your data is not very accurate:

For a 2-port antenna, CRSs consume 9.5% of air interface resources. Thisoverhead increases to 14.3% for a 4-port transmission configuration. Check here : https://www-file.huawei.com/-/media/CORPORATE/PDF/white%20paper/tm9-white-paper.pdf?la=en

[electron]

Hello friends,

Please note that deploying MIMO 4x4 in the network is not all about penetration of 4x4 support devices . When there is migration from 2x2 to 4x4 there is huge gain in spectral efficiency not only due to supportive 4x4 devices. It is mostly about diversity gain we achieve in both DL and UL which is directly interpreted to ~20% capacity gain. Aside the higher cost for operators caused by new RRU installment and other SW packages they need to purchase , Interference to other cells may increase for surrounding sites that may impact their efficiency.

B
R

[subway]

Hello friends,

... It is mostly about diversity gain we achieve in both DL and UL which is directly interpreted to ~20% capacity gain...
B
R

What you say about diversity is generally true, but has nothing to do with MIMO mode directly. For example if you have an 8T8R TD-LTE system (like we do), you can configure it as the following:

The cell is in 8T8R mode: 8R diversity is enabled for the uplink, and 8T is enabled for the downlink, but we are still on CRS2 (2x2 MIMO). In order to utilize diversity, you dont need to use higher order MIMO mode. You just need enough active antenna elements, and th relevant licenses for 4 or 8T8R diversity.

On 8T8R systems (especially if its not dense urban), the wise thing to do is TM9 with MUBF, this way you can selectively utilize the 4 spacial layers for capable UEs (and radio conditions) without the massive fixed CRS overhead of the classical 4x4 MIMO (TM3/TM4). If there are no TM9 capable UEs, keep it on CRS2, and use 4T4R or 8T8R in diversity mode, not in 4x4 MIMO mode.

[kaikunika]

the disadvantage is the cost . i dont think there will be a power problem even if you used fixed RRU

[electron]

What you say about diversity is generally true, but has nothing to do with MIMO mode directly. For example if you have an 8T8R TD-LTE system (like we do), you can configure it as the following:

The cell is in 8T8R mode: 8R diversity is enabled for the uplink, and 8T is enabled for the downlink, but we are still on CRS2 (2x2 MIMO). In order to utilize diversity, you dont need to use higher order MIMO mode. You just need enough active antenna elements, and th relevant licenses for 4 or 8T8R diversity.

On 8T8R systems (especially if its not dense urban), the wise thing to do is TM9 with MUBF, this way you can selectively utilize the 4 spacial layers for capable UEs (and radio conditions) without the massive fixed CRS overhead of the classical 4x4 MIMO (TM3/TM4). If there are no TM9 capable UEs, keep it on CRS2, and use 4T4R or 8T8R in diversity mode, not in 4x4 MIMO mode.

Well, It absolutely has correlation though never talked about MIMO mode diversity! Tell me how you can justify the gain of deploying MIMO 4x4 for the market which has negligible penetration of supportive 4x4 devices but you still see about 20% gain in efficiency? No doubt MIMO diversity gain for both DL and UL will reflect to spectral efficiency and higher capacity gain

B
R

[auto_art]

Also with 4t4r signalling overhead is increased( more blocks to reference) in our case we observed reduced throughput. Only single type of ue in wttx setup for our network.

[subway]

That is the point: I cant justify it, and neither most of the operators, that is the reason you see so little amount of 4x4 MIMO configured networks. The ones you see are also above 2GHz (even 2100MHz networks do not see 4x4 configurations). I tested 4x4 MIMO (TM3 and TM4) in our network quite a lot where most clients are 4x4 capable. Yet the gains (in terms of spatial multiplexing, not in terms of diversity gain) were very limited, only close by CPEs were able to switch higher than Rank2, but even in this case it was mostly Rank3 and not Rank4, this was due to the non-reflective rural environment. Even CPEs with 23-25dB SINR were not able to swtich to higher than Rank2 in most cases, and when I forced them to do so, the performance was worse then on CRS2. And for the 5-10% of CPEs that can switch to Rank3, all the CPEs are loosing capacity doe to the higher fixed CRS overhead: even lab tests indicated a drop from 110Mbit RLC DL speed to about 85Mbit RLC DL speed for Rank2 CPEs (per 20MHz carrier). So Huawei can say the extra overhead of CRS is only 9.5 --> 14%, but the drop the UEs suffer are definitely more than that.

I dont want to convince anybody, non the less my own experience is simply not very good about 4x4 MIMO itself (which is not to be mistaken by 4T4R or 8T8R antenna and cell configurations). It is also quite clear that the industry is not forcing classical 4x4 MIMO, instead BF, multi user MIMO and advanced IRC is the way to go as they yield much better gains and less fixed overhead.

[electron]

That is the point: I cant justify it, and neither most of the operators, that is the reason you see so little amount of 4x4 MIMO configured networks. The ones you see are also above 2GHz (even 2100MHz networks do not see 4x4 configurations). I tested 4x4 MIMO (TM3 and TM4) in our network quite a lot where most clients are 4x4 capable. Yet the gains (in terms of spatial multiplexing, not in terms of diversity gain) were very limited, only close by CPEs were able to switch higher than Rank2, but even in this case it was mostly Rank3 and not Rank4, this was due to the non-reflective rural environment. Even CPEs with 23-25dB SINR were not able to swtich to higher than Rank2 in most cases, and when I forced them to do so, the performance was worse then on CRS2. And for the 5-10% of CPEs that can switch to Rank3, all the CPEs are loosing capacity doe to the higher fixed CRS overhead: even lab tests indicated a drop from 110Mbit RLC DL speed to about 85Mbit RLC DL speed for Rank2 CPEs (per 20MHz carrier). So Huawei can say the extra overhead of CRS is only 9.5 --> 14%, but the drop the UEs suffer are definitely more than that.

I dont want to convince anybody, non the less my own experience is simply not very good about 4x4 MIMO itself (which is not to be mistaken by 4T4R or 8T8R antenna and cell configurations). It is also quite clear that the industry is not forcing classical 4x4 MIMO, instead BF, multi user MIMO and advanced IRC is the way to go as they yield much better gains and less fixed overhead.

Just to correct you about the base of your logic which mislead is that there are many operators investing on 4x4 MIMO if they are very much focus on spectral efficiency!

I am personally giving strategic RAN consultation to one of the most largest and most loaded LTE network in the world and for your information we have 40% of network running on MIMO 4x4 and if there is no budget constraint more MIMO 4x4 will be deployed . Below graphs are based on live network statistics which i prepared . Might be another proof point for you though we have very negligible supportive 4x4 devices in the network !

42837



B
R

[subway]

Looking at your graphs I see no contradiction about what we discussed so far: switching from 2T to 4T will increase the downlink diversity by 3dB, using 4R is also adding about 3dB gain (which is even more needed on the uplink then on the downlink side). Combine that with smart IRC: this obviously increases spectral efficiency along with channel quality for every customer, 2x2 capable UEs as well. But not because of extra spatial layers: mainly diversity and secondly smart combining (not sure if you use it). Maybe I did not highlighted this before and this is the reason we misunderstand each other a bit. I think it would be better to call it 4T4R and not 4x4 MIMO, as quite a lot of people are using these as synonyms (incorrectly). Or even call it CRS2 and CRS4.

What is interesting in your charts is the average cell Tput combined with the information that there is a low number of 4x4 capable UEs. That indicates - if your 4T4R network is configured for CRS4 (is it?) - that the diversity gain is outperforming the CRS overhead.

"40% of network running on MIMO 4x4"

Hm.. 40% of what? Traffic, or sites? TDD or FDD? On which frequencies are 4T4R being deployed? I think I have seen no major antenna vendors having 4T4R capability below 2.1GHz where most traffic (at least in Europe) is transmitted, mostly due to physical constrains on the antenna side (also checked some AAUs). So hearing 40% of one of the largest and most busy 4G network is on 4T4R definitely made me interested. I think we can learn a lot from this, but a bit more information is needed.

[electron]

Looking at your graphs I see no contradiction about what we discussed so far: switching from 2T to 4T will increase the downlink diversity by 3dB, using 4R is also adding about 3dB gain (which is even more needed on the uplink then on the downlink side). Combine that with smart IRC: this obviously increases spectral efficiency along with channel quality for every customer, 2x2 capable UEs as well. But not because of extra spatial layers: mainly diversity and secondly smart combining (not sure if you use it). Maybe I did not highlighted this before and this is the reason we misunderstand each other a bit. I think it would be better to call it 4T4R and not 4x4 MIMO, as quite a lot of people are using these as synonyms (incorrectly). Or even call it CRS2 and CRS4.

What is interesting in your charts is the average cell Tput combined with the information that there is a low number of 4x4 capable UEs. That indicates - if your 4T4R network is configured for CRS4 (is it?) - that the diversity gain is outperforming the CRS overhead.

"40% of network running on MIMO 4x4"

Hm.. 40% of what? Traffic, or sites? TDD or FDD? On which frequencies are 4T4R being deployed? I think I have seen no major antenna vendors having 4T4R capability below 2.1GHz where most traffic (at least in Europe) is transmitted, mostly due to physical constrains on the antenna side (also checked some AAUs). So hearing 40% of one of the largest and most busy 4G network is on 4T4R definitely made me interested. I think we can learn a lot from this, but a bit more information is needed.

Hi,

I guess now you fully got my points !

Note : 40% of the sites are configured and working with MIMO 4x4 in my network


B
R

[subway]

To provide some more insight, I have reconstructed a site from 8T8R CRS2 to 8T8R CRS4, (all CPEs are 4x4 capable), and the results are the following:

1. An instant drop of RSRP of an average of 4.2dB, in some cases it reached as high as 5.6dB (the expected degradation is 3dB).
2. An instant drop of the average DL SINR by 2.9dB, which scales up by traffic and in busy hours it can get as worse as 5dB degradation.
3. A drop in RSRQ, also scales up in peak hours.
4. The proportion of Rank3 or Rank4 combined is around 10-15%, from this it is no surprise that,
5. The average traffic dropped by 20% (100% PRB utilization before and after the reconstruction).
6. We dont have any mobility, but if we would, I am sure we see an increased drop call rate based on the above results.

So, based on the results it is clear that 4x4 MIMO (CRS4 to be clear) is not suitable for rural coverage, even with fixed high gained CPEs, and it is not giving an instant gain in every circumstances automatically.

The site will be reconstructed to CRS2, and the extra antenna chains will be utilized for beamforming, as that provides better gains than 4x4 MIMO.

[justdream]

Just to correct you about the base of your logic which mislead is that there are many operators investing on 4x4 MIMO if they are very much focus on spectral efficiency!

I am personally giving strategic RAN consultation to one of the most largest and most loaded LTE network in the world and for your information we have 40% of network running on MIMO 4x4 and if there is no budget constraint more MIMO 4x4 will be deployed . Below graphs are based on live network statistics which i prepared . Might be another proof point for you though we have very negligible supportive 4x4 devices in the network !

42837




B
R
Friend, is it possible share the document has this MIMO comparison

[NPERF]

Can any one can share this document for a review and study - thanks

[mysterm]

effectively, a degraded coverage with more throughput for the same fixed power..