Worst-case Optimized V-BLAST Receiver Design for Imperfect MIMO Channels

V-BLAST systems require perfect knowledge of the channel at the receiver, which could be impractical in real world. The channel state information (CSI) that contains the unknown estimation error will hurdle the zero forcing based intersymbol interference nulling and symbol detection ordering in V-BLAST. This paper extends the V-BLAST scheme to the imperfect CSI under the assumption that the channel estimation errors are unknown and norm-bounded. The extension consists of novel worst-case optimized zero forcing nulling and symbol detection ordering. In each layer, the zero-forcing nulling is designed to minimize the worst case error power due to the imperfect channel estimation, while the order of symbol detection is organized to detect the symbol corresponding to the best worst-case postprocessing signal over interference plus noise ratio (SINR). We have shown in computer simulations that the proposed algorithm considerably outperforms the standard ZF V-BLAST in the worst case CSI, while making no significant degradation in non worst-case scenarios