Performance of a blind multistage decision feedback equalizer for MIMO systems

In this work, we consider the problem of blind source separation of QAM sources transmitted over a broadband MIMO system, by investigating the effect of adding a decision feedback equalizer (DFE) to our previously proposed blind multistage equalization scheme. A modified Constant Modulus Algorithm (CMA), which includes the Alphabet Matched Algorithm (AMA) has been shown to successfully equalize and separate multiple data blocks transmitted over an IID Rayleigh fading channel. A decision feedback equalizer (DFE) is added to the multistage equalization scheme, and a new channel estimator is proposed which incorporates the feedback equalizer. At each stage of equalization, the channel estimator provides an estimate of a single column of the MIMO channel (including all taps). The channel estimator based on DFE is shown to provide a more accurate estimate of the channel vector, and generally improves the overall performance of the multistage scheme. The performance of the multistage decision feedback equalizer and its corresponding channel estimation scheme is evaluated compared to the previously proposed multistage CMA+AMA equalization and signal separation method, by considering the symbol error rates (SER) of the captured sources at different stages, as well as the channel estimation errors.