The popular Alamouti orthogonal Space Time Block Code (STBC) attains full transmit diversity in two transmitter multiple input multiple output channel systems. Maximum ratio combiner (MRC) performance depends mainly on the quality of channel estimator, which depends mainly on the number of training symbols. The aim of this paper is to improve performance of MRC of QPSK-Alamouti’s STBC's decoder without increasing number of training symbols. This paper gives an introduction to the basic concepts of training based channel estimator and explains the implementation of least square error (LS) channel estimator with diagonal and orthogonal training matrix. The kurtosis based source extraction method based on using real imaginary (R-Im) decomposition of MRC was fully described. Finally the benefit of using at least four training symbols for initialization de-mixing vector and removing source ambiguity was illustrated.
Computer simulation for QPSK Alamouti STBC's in flat fading MIMO channel was implemented using MATLAB2012. First MRC decoder with LS channel estimator technique analyzes according to their number of training symbols (Nt=2,4,…10) and type of training matrix (diagonal or orthogonal ). We found that: orthogonal training matrix for any sequence length provides superior performance than diagonal training matrix. Finally the proposed decoding technique was implemented and it's BER performance were analyzed using only four training symbol with illustration for number of iteration at each SNR.