A semi-blind detection algorithm for V-BLAST system

The channel information is required in conventional V-BLAST algorithm. When the training sequences are short, the channel estimation is very noisy, which incurs the system performance degradation. We propose a semi-blind detection algorithm in this paper. Signals transmitted from different antennas are separated blindly first based on the statistical independence of the signals. Short training sequences are then utilized to eliminate the permutation and scale ambiguities which are inherent to blind separation algorithms. Simulation results demonstrate that the performance of the proposed algorithm is better than that of the conventional V-BLAST algorithm both in flat and frequency-selective fading channel when the training sequences are short.