SIR-optimized weighted linear parallel interference cancellers on Rayleigh fading channels

In this paper, we present weighted linear parallel interference cancellers (LPIC) where the estimated multiple access interference (MAI) from each interfering user is weighted by a factor before cancellation on Rayleigh fading and diversity channels. We obtain exact expressions for the signal-to-interference ratio (SIR) at the output of the cancellation stages which we maximize to obtain the optimum weights for each interfering user in each cancellation stage. For a system that uses same weight for all interfering users in a given stage, we obtain exact closed-form expressions for the optimum weights for the different stages of the LPIC. We show that this SIR-optimized weighted LPIC scheme clearly outperforms both the matched filter (MF) detector as well as the conventional LPIC (where the weight is taken to be unity for all interfering users and for all stages), in both near-far as well as non-near-far conditions.