Why does ltePMISelect select PMI using summed SINR?

In the source code of ltePMISelect (\toolbox\lte\lte\ltePMISelect.m), I found the PMI are selected based on maximum summed SINR criteria as shown in below. It seems makes more sense to select PMI based on maximum summed capacity given by sum(log2(1+SINR)). Can someone from the team developed this explain and point to the reference paper if possible?

      if (wideband_i1)
          % wideband part
          totalSINR=sum(sum(SINRs,1),2);
          totalSINR=reshape(totalSINR,N1,N2);
          best1=find(max(totalSINR,[],2)==max(max(totalSINR,[],2)));
          if (~isempty(best1))
              best1=best1(1);
          end
      end

Thanks a lot in advance.

Answer 1

[Iain Stirling]

Hello Chulong,

The selection criteria used is based on the linear MMSE SINR metric assumed in the design of the codebook entries during the standardisation process. This was chosen not solely as a function of the theoretical performance but also with regard to the implementation complexity and feedback quantization. The following paper gives a comparison of various techniques: D. J. Love and R. W. Heath, “Limited Feedback Unitary Precoding for Spatial Multiplexing Systems,” IEEE Trans. Inf. Theory, vol. 51, no. 8, pp.2967–2976, Aug. 2005.

Regards,

Iain.