Application of a blind channel estimator for source separation in MIMO systems

The performance of a blind channel estimator is evaluated in this work. This channel estimator approximates the channel impulse responses in Multiple-Input Multiple-Output (MIMO) communication links without any prior knowledge of either the transmitting channel or the transmitted signals. It uses the autocorrelation of signals at the receiver end.In this paper as a major application of this channel estimation scheme, it is applied to achieve blind source separation (BSS) in a multiple antenna wireless communication system. The performance of the blind channel estimation scheme is compared using an improved version of the CMA equalizer for QAM signals [5] with the training-based Minimum Mean Squared Error (MMSE) and Maximum Likelihood (ML) channel estimators. The frequency and impulse response of the estimated channels are compared to those of the actual channels, and the channel estimation errors are plotted over a range of Signal to Noise Ratio (SNR) for each of the three channel estimators.