Adaptive Blind Equalization with Multichannel CRIMNO Algorithm

A new adaptive blind equalization algorithm for multichannel systems is presented in this paper. The cost function is an augmented version of CRIMNO (criterion with memory nonlinearity), which also includes the mean-square error criterion between the outputs of the multichannel equalizers. It is shown that by forcing the outputs of the equalizers to be equal, as should be during perfect equalization, the convergence of the MCRIMNO (Multichannel CRIMNO) technique is faster than its individual single-channel versions. The convergence of the new algorithm is independent of time or frequency shifts of the received sequences. Simulation results clearly demonstrate the improved performance of the MCRIMNO technique.