Channel state information based successive interference cancellation as well as quality of service based successive interference cancellation are unable to remove the error floor in uplink NOMA. So, how the error floor can be removed?
Do you mean that the BER converges to a limit as the SNR goes to infinity?
If you have a single receive antenna, you only have one spatial degree of freedom so you cannot get a multiplexing gain that is larger than one whatever you do. So if you are trying to serve multiple users and apply successive interference cancelation, an error floor is unavoidable.
One solution is to turn off the NOMA functionality when it becomes suboptimal; i.e., switch to OMA. I think the "rate splitting" framework is doing that in an optimized manner.
Another solution is to equip the receiver with multiple antennas. Then you are entering the field of multi-user MIMO, which is the extension of NOMA that is considered in 5G.
Hi Anand Jee
The error floor in uplink is always there regardless of the CSI. It is proved in Article Error Probability Analysis of NOMA-based Diamond Relaying Network
. Nervetheless, with a joint ML decoding, the error performance can be improvedArticle BER Performance of Uplink NOMA with Joint Maximum-Likelihood Detector
. On the other hand, a recent paper discuss the error floor in terms of outage probabilityPreprint Unveiling the Importance of SIC in NOMA Systems: Part I -- S...
Regards
Hi Anand,
Error floor in uplink NOMA will be there irrespective of channel state information. Please refer to the Section 3B for further details in this manuscript:
Article Error Probability Analysis of NOMA-based Diamond Relaying Network
However, apart from channel coding, designing spreading sequences play significant role in error performance of uplink NOMA systems. If compressed sensing based multi-user detection is used, then sequences with low mutual coherence might provide improved error performance. Regards!
Anand Jee
Hi Anand,
I guess that channel coding alone will not be able to change the diversity order, so the error-floor can be not be avoided. However, using a multiple-antenna receiver with channel coding may result in a residual error within acceptable limits.
Another dimension to explore in this scenario can be some well-designed modulation scheme that may avoid the error floor. Although this will not be simple, I have (sometimes ago) seen a paper where the authors claimed that their constellation can mitigate the problem of error propagation due to imperfect SIC.
Hi Anand,
The error floor in uplink can be removed if you change the SIC receiver to a joint maximum likelihood one, which is feasible in uplink, as mentioned by Mohammad Ahmad Al-Jarrah
.Firstly, note that for let say a two-user uplink NOMA system with SIC receiver at the BS, when the BS tries to recover the high power user first, it treats the signal of other user as noise, which causes an error floor. This error floor will always exist in SIC based receiver due to this "treating as noise" of the low power users.
A solution to this is to perform a joint maximum likelihood reception. I read a paper a couple of months ago titled "BER Performance of Uplink NOMA With Joint
Maximum-Likelihood Detector", as mentioned above too, with the following link
Article BER Performance of Uplink NOMA with Joint Maximum-Likelihood Detector
They have performed the mathematical analysis and have shown clearly how the error floor is removed (Figs 3 and 4 in the paper). The performance can vary based on the number of receive antennas, but the error floor can still be avoided. This is because, in joint maximum likelihood, you are jointly recovering the signals of both users (no one is treated as noise), and therefore a better performance with no error floor.
The authors were kind enough to send me the MATLAB code for the analytical analysis (which I needed), but I can't share with you as it is theirs. But, if you send them an email, they will surely send it to you too.
I hope this answers your query.
Kind Regards,
Basit
Hi Anand Jee
Perhaps the following two references might Help!
A. A. Khansa, Y. Yin, G. Gui and H. Sari, "Power-Domain NOMA or NOMA-2000?," 2019 25th Asia-Pacific Conference on Communications (APCC), Ho Chi Minh City, Vietnam, 2019, pp. 336-341, doi: 10.1109/APCC47188.2019.9026468.
Article Performance analysis of Power-Domain NOMA and NOMA-2000 on A...
Good luck!
- Badges
- Science method