Adaptive detection of V-BLAST in slowly time-varying channels

V-BLAST is able to achieve high spectral efficiency by spatially multiplexing independent data streams. However, the way to demultiplex the multiplexed signals is a nontrivial work, especially when the channel is continuously time-varying. In this paper, we study how the channel variation affects the performance of the zero-forcing based nulling (ZFN), which is a key operation in the prototype V-BLAST detection method (Golden et al. (1999)). The study shows that the performance of ZFN degrades very soon when the nulling vectors are outdated, yet the direct updating of the nulling vectors is of very high computation complexity. To deal with this problem, we imitate some ideas from adaptive multiuser detection for CDMA and propose an adaptive detection method. We discuss some implementation aspects of the proposed method and support our development with computer simulations, which show that the proposed method can be used when modest performance is required under strict implementation complexity constraint.