Treating Self-Interference as Source: An ICA Assisted Full-Duplex Relay System

Abstract

We investigate an amplify-and-forward (AF) full-duplex (FD) relay system, where the FD incurred self-interference (SI), through partial cancelation at relay, is treated as a useful source at destination to enhance degree of freedom in signal detection, while reducing the signal processing cost of SI cancelation. An independent component analysis (ICA) based equalization structure is employed at destination to separate and detect the desired signal from the residual SI in a semi-blind way. The mode of SI cancelation at relay is chosen adaptively based on the threshold of signal-to-interference ratio (SIR) at relay. The proposed FD relay system not only features reduced signal processing cost of SI cancelation, but also achieves much higher energy efficiency (EE) than conventional FD relay systems where SI is canceled as much as possible. Also, the proposed system enables full resource utilization via consecutive data transmission at all time and the same frequency, leading to much higher throughput and EE than the conventional time-splitting and power-splitting based SI recycling approaches that occupy partial resources. Last but not least, the proposed system demonstrates a bit error rate (BER) performance that is robust against a wide range of SI and close to the ideal case with perfect channel state information (CSI) and perfect SI cancelation, while requiring no training sequence for estimation of any channel involved.