Thanks Thanks:  1
Showing results 1 to 9 of 9

Thread: TM8 vs. TM9

  1. #1
    Member Reputation: 108
    Join Date
    2015-01-07
    Posts
    136


    Default TM8 vs. TM9

    We just introduced TM9 (2 and 4 layers) in our TDD network, and I noticed that the TM9 scheduled user number is still relatively low and quite a lot of the CPEs are still on TM8. All CPEs support TM9 4 layer and CSI-RS 8, and this is how the network is configured. The busiest sector has around 70-80 CPEs and only 5-10 is scheduled in TM9 mode. Most of the CPEs are evenly distributed in the coverage area, with line of site. If someone have any ideas, please let me know. We use Huawei eNodeBs and SRAN12.1

    MOD1:

    The Huawei implementation guide tells that TM9 requires Single Stream beamforming to be set, but say nothing about what to do with Dual-Stream beamforming when all the UEs are TM9 capable? I admit it is likely an unrealistic scenario in most of the networks, but maybe the reason I see very little TM9 activity is because I should disable Dual-Stream beamforming, as the ones capable of doing TM8 Rank2 can also do TM9 Rank2, and TM8 Rank1 can be switched to TM7. But this is just a theory.

    MOD2: so I tested the above theory and switched off dual stream BF. The result is disappointing: the TM9 avg. user number did not increased, TM8 is gone, TM7 is still zero. So where does the traffic go? Most of it to TM3, small number of it to TM2.

    It is also interesting that the "CellBfMimoParaCfg.Tm3AndTm9ThdOffset" value is controlling the TM3 to TM9 switching, and the TM9 implementation guide recommends to set this to the lowest possible value (-30), while the weblmt recommends this to be set to 0. I think I will set this back to the weblmt recommended value and give it a try.
    Last edited by subway; 2020-02-05 at 10:07 PM

  2. # ADS
    Circuit advertisement
    Join Date
    Always
    Posts
    Many
     

  3. #2
    Member Reputation: 238
    Join Date
    2009-01-20
    Posts
    539


    Default Re: TM8 vs. TM9

    Hi

    In network we have deployed TM7, when asked about TM8 -said it has yielded low results.

    Any observation on TM7 and TM8 deployment.

    Vendor : huawei

    Regards
    Plannerguy

  4. #3
    Member Reputation: 108
    Join Date
    2015-01-07
    Posts
    136


    Default Re: TM8 vs. TM9

    Hi! Thanks for the answer!

    Our sites and CPEs are all TM8 capable, and there are enodebs where all the CPEs are TM9 capable. As these are all static, I figured that the best option would be to use TM9 with PMI feedback. The core problem is that most of the CPEs did not switched from TM8 to TM9 adapitvely. I finetuned some parameters, but still the TM9 traffic was not increased. So yesterday I forced a complete site to TM9wPMI mode (all CPEs are TM9 capable), so lets see what will happen in terms of speed and PRB utilization. This network is also Huawei on both the RAN and CPE side, but the one thing we will never buy again is Huawei CPEs...

  5. #4
    Member Reputation: 79
    Join Date
    2010-05-15
    Posts
    104


    Default Re: TM8 vs. TM9

    Dear Subway,

    Finally, What is the best configuration for TM9? Can you please share the script to activate TM9 for TDD?
    What about MU-Beamforming have you tested it in your network?

    Thank you in advance,

  6. #5
    Member Reputation: 108
    Join Date
    2015-01-07
    Posts
    136


    1 out of 1 members found this post helpful.

    Default Re: TM8 vs. TM9

    It is in the plan to test MU-BF, but we are not there yet. Hopefully the extra traffic we see now will push the management to invest in it.

    On SRAN12 the TM9 feature is a waste of money, we gained less then 10% about it --> not worth it. And the CPEs are all TM9 capable, so we have a 90+% of TM9 capable clients. I suspect - based on the release notes - that between SRAN12 and SRAN15, there was substantial development and bugfixing done to TM9 and BF in general, but I can't confirm it until the upgrade is done.

    So it is bit inconclusive...

  7. #6
    Member Reputation: 79
    Join Date
    2010-05-15
    Posts
    104


    1 out of 1 members found this post helpful.

    Default Re: TM8 vs. TM9

    Dear subway,

    Thank u very much for your answer.We are on sran13 we will launch the trial in the coming days , also all the CPEs are TM9 capable in our network, I will let u know about the results

  8. #7
    Member Reputation: 108
    Join Date
    2015-01-07
    Posts
    136


    Default Re: TM8 vs. TM9

    Hope your CPEs are not from Huawei like ours...

    I trialed multi.user BF and TM9 in tandem, and the result was very minimal. Barely any gain at all on heavily loaded sites (PRB usage is 90%+ in busy hours). One thing I noticed though is that despite the measured SINR did not improved at all, the CQI in the same time went from 11-12 to 14-15. In the same time: no noticable PRB utilization decrease, and no more DL.bits carried. And the two features combined should give a huge boost for a rural setup...

    tm9+mubf.jpg

  9. Thanks justdream thanked for this post
  10. #8
    Member Reputation: 108
    Join Date
    2015-01-07
    Posts
    136


    Default Re: TM8 vs. TM9

    I am still thinking about how can it happen, that the CQI improves quite a lot, and compared to that the SINR does not change at all, or slightly degrades. As the CQI is one of the least properly defined performance indicator, and there is also a large amount of difference between modem vendors, I would say this is another crap the Huawei balong modem generates in TM9 mode, but as I have no other CPE type to compare with... The one thing TM9 should improve is the SINR, as it forms the PDSCH beam towards the CPEs. And yet, that is not the case, or at least the modem reports something else. And the above picture is from a CPE in the middle of the primary beam, no intra-interference.

    If anyone has any theory, I am happy to hear it :-)

  11. #9
    Member Reputation: 156
    Join Date
    2010-03-26
    Posts
    74


    Default Re: TM8 vs. TM9

    Hi,

    I want to know if you have 4T4R sites or 8T8R sites and whether you have CA or not?

    Generally CQI is lower with CA compared to non-CA. MUBF and CA don't work well together unless there is UL-CA supported CPEs and UL-CA is enabled in your network.
    So, if you enabled MUBF and CA usage reduced, your CQI can improve.

    Secondly, in beamforming modes TM7/TM8/TM9, RSRQ degrades while SINR doesn't. This is due to improved coverage of beamforming. Perhaps CQI gives more weight to coverage than SINR and other factors, so it shows improvement with beamforming.

    Thanks.

Bookmarks

Bookmarks

Posting Rules

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •