Decision feedback equalizer with the blind matched filter estimation

The optimal receiver front-end is the matched filter (MF) for a maximum SNR in a communication channel with AWGN. The sampled MF output can be processed with the suboptimal yet computationally effective decision feedback equalizer (DFE). In this paper, the filters of the DFE are derived in terms of the channel autocorrelation sequence which is estimated blindly using the sampled channel output signal. Initially, the structure is operated in the linear equalizer (LE) mode (without the decision device (DD)) with the symbol-rate sampled MF placed at the end. During this LE mode, the MF is estimated blindly using the CMA. After it converges, the structure is put back into its original form and the DD is included in the feedback filter (FBF). The MF is adapted in the system identification structure using the detected symbols from the DD. In this way, the MF estimate from the CMA is improved further and an all-blind DFE is realized for a SISO system. The BER values obtained from this structure are very close to those obtained with the true values of the MF and hence they prove that the structure is efficient and effective.