Characterizing Achievable Rates for Two-Path Digital Relaying

This paper studies bandwidth-efficient transmission protocol for a two-hop communication network with one source, one destination, and two assisting relays. The two-path successive relaying protocol is considered where the source/destination transmits/receives continuously in time, and two half-duplex relays both operating in a time-division-duplex (TDD) mode successively in turn forward messages from source to destination. By mimicking a full-duplex relay with two alternating half-duplex relays, this protocol ideally achieves the same bandwidth efficiency. However, a major issue for this protocol to be effectively implemented is to tackle the inter-relay interference (IRI), which occurs when one relay listens to the source, it is interfered with by the other relay that transmits to the destination. This paper characterizes achievable rates for the protocol of interest for a simplified symmetric two-path relay channel when both relays employ the decode-and-forward (DF) operation or digital relaying. First, the achievable rate of digital relaying using the conventional repetition coding is derived, and is compared to that of a full-duplex relay. It is shown that owing to the IRI, this method leads to severe rate loss at high signal-to-noise- ratio (SNR) regime. Then, this paper presents a new scheme that can be easily incorporated into the considered protocol to avoid completely any rate loss caused by the IRI. The proposed scheme utilizes the dirty-paper-coding (DPC) -based interference pre-subtraction at the source transmitter, and is shown to be capacity-optimal.