Superimposed XOR: Approaching Capacity Bounds of the Two-Way Relay Channels

TL;DR

This paper introduces superimposed XOR, a novel physical layer network coding scheme for two-way relay channels, which effectively handles channel and information asymmetries to approach capacity bounds.

Contribution

The paper proposes a new superimposed XOR scheme that improves rate regions and performance over existing XOR and superposition coding in asymmetric two-way relay channels.

Findings

Achieves larger rate region than XOR and superposition schemes.

Performs significantly better over fading channels.

Gap to capacity bound approaches zero at high SNR.

Abstract

In two-way relay channels, bitwise XOR and symbol-level superposition coding are two popular network-coding based relaying schemes. However, neither of them can approach the capacity bound when the channels in the broadcast phase are asymmetric. In this paper, we present a new physical layer network coding (PLNC) scheme, called \emph{superimposed XOR}. The new scheme advances the existing schemes by specifically taking into account the channel asymmetry as well as information asymmetry in the broadcast phase. We obtain its achievable rate regions over Gaussian channels when integrated with two known time control protocols in two-way relaying. We also demonstrate their average maximum sum-rates and service delay performances over fading channels. Numerical results show that the proposed superimposed XOR achieves a larger rate region than both XOR and superposition and performs much better over fading channels. We further deduce the boundary of its achievable rate region of the broadcast phase in an explicit and analytical expression. Based on these results, we then show that the gap to the capacity bound approaches zero at high signal-to-noise ratio.