Conditional Decoding for X Channels

Abstract

The conditional decoder is a low complexity and optimal decoder for multiplexed orthogonal designs in point-to-point channels. We extend this notion to X channels and differentiate between two interference scenarios. First, the interference is perfectly aligned in a different sub-space than the intended information symbols. For this scenario, the conditional decoder is applied as a one stage and two stages decoder to cancel the interference and decode the desired signal. Second, the interference is misaligned. In this case, the conditional decoder attempts to jointly cancel the interference and decode the intended signal while achieving a performance gain over other interference cancellation schemes. We consider the two user scenario where Alamouti codes are used at the transmitters and then extend our investigation to three user channels with arbitrary signal structure. Our numerical results establish the superiority of our decoder to previously proposed zero-forcing and decoupling techniques, in terms of performance. It is further shown that the proposed decoder achieves the same performance as the sphere decoder; while enjoying a much lower implementation complexity.