Multi-user multimodulus algorithm in blind source separation and equalization for MIMO systems

Abstract

In frequency selective channels, the multiple-input multiple-output (MIMO) system introduces both inter-symbol interference (ISI) and inter-source interference to the receiver. Existing algorithms, such as the multi-user constant modulus algorithm (MU-CMA), suffer from steady state phase errors (if phase-locked loop is not used) and slow speed in symbol recovery from the above interferences. We propose a new blind source separation and equalization algorithm based on the multimodulus algorithm (MMA) and a cross-correlation function of all equalizer outputs to overcome the above problems. We will refer to our new algorithm as MU-MMA. The results show that MU-MMA can compensate the phase error and converge faster than MU-CMA.