Efficient Square-root Algorithms for the Extended V-BLAST with Selective Per-Antenna Rate Control

摘要

We propose efficient algorithms for the extended Vertical Bell Labs Layered Space-Time architecture (V-BLAST) with selective per-antenna rate control (S-PARC). Instead of computing SNIRs from nulling vectors, we compute SNIRs from diagonal entries in the estimation error covariance matrix P, which are obtained from entries in F, i.e. the triangular square-root of P. Then the square-root V-BLAST algorithm is applied, to compute F for an antenna subset from F for another subset. Moreover, when computing F, we can reuse intermediate results to further reduce the computational complexity dramatically. Assume M transmit/receive antennas. The proposed algorithm for the S-PARC scheme minimizing total transmit power has the speedup of 0.14M + 1.82, with respect to the corresponding S-PARC algorithm computing SNIRs from nulling vectors, while the proposed algorithm for the S-PARC scheme maximizing total information rate has the speedup of (M + 58)/24.