Joint Encoding and Grouping Multiple Node Pairs for Physical-Layer Network Coding With Low-Complexity Algorithm

Abstract

How to put fifth-generation (5G) into the commercial service by 2020 is still challenging. Since physical-layer network coding (PNC) does not bring unnecessary redundancy and offers satisfactory end-to-end services, it is promising to provide network services in 5G networks. Existing works on PNC mainly focus on the case that two packets are simultaneously sent by a node pair for information exchange, and the transmitted signals are superposed and encoded at a relay. To further exploit the potential performance improvement of PNC, this study aims at encoding multiple packets through jointly processing multiple superposed signals sent by diverse node pairs. Particularly, we formulate the problem of joint-encoding method design as a binary integer programming model and propose a low-complexity solution based on the idea of filling a Latin rectangle. We find that network throughput enhancement does not accompany the increasing number of coded node pairs, and some node pairs cannot be encoded together since they fail to decode the resulting encoded packet. Therefore, the decodability-based node pair grouping scheme is further investigated. Simulation results in various scenarios show that the proposed low-complexity scheme performs neck to neck to the optimum one with high computational complexity.