GA Based Estimation of Sparse MIMO Channels with Superimposed Training
Multiple-input multiple-output (MIMO) techniques are foreseen to play a vital role in future 5G cellular networks. This paper presents a novel approach that employs genetic algorithm (GA) to estimate the sparse uplink MIMO channels using superimposed training sequence (SiT). At each transmitter (user) a training sequence is mathematically added with the data bits; thus, avoiding the overhead of dedicated frequency/time slots used for the training. On the receiver side (base station), signals received at all the receive antennas are jointly processed by employing the proposed method to obtain channels’ estimate. Then, a linear minimum mean square error (LMMSE) equalizer estimates the data sequences sent by transmitter. A computer simulation based performance analysis of the proposed method is presented, where performance evaluation is done using metrics of normalized channel mean square error (NCMSE), as well as, bit error rate (BER). A comparative analysis of the proposed method with notable SiT least squares (SiT-LS) and SiT-LMMSE methods in the literature is conducted, which clearly demonstrates that the proposed method outperforms both the existing techniques.