A Framework for Evaluating Physical-Layer Network Coding Gains in Multi-Hop Wireless Networks

Physical-Layer Network Coding (PLNC) was first introduced as a solution to increase the throughput of a two-way relay channel communication. Unlike most wireless communication techniques which try to avoid collisions, PLNC allows two simultaneous transmissions to a common receiver. In basic topologies, this technique has been shown to significantly enhance the throughput performance compared to classical interference-free communications. However, quantifying the impact of PLNC in large multi-hop networks remains an open question. In this work, we introduce the first theoretical framework that, given a particular network topology and traffic matrix, can evaluate the optimal network throughput subject to a fairness constraint relative to the initial demand of each flow, when PLNC is adopted. Based on linear programming, our solution ensures to respect the particularities imposed by PLNC. We use this framework to evaluate three state-of-the-art PLNC schemes across a variety of topologies and traffic matrices. Our numerical analysis reveals that while in simple toy topologies PLNC can significantly increase the network throughput, in large topologies the verdict is mixed. For certain topologies and traffic patterns adopting PLNC can double the throughput while in others, depending on the scheme used for implementing PLNC, the gain can be as high as 60 percent or as low as 0 percent when compared to traditional interference-free transmissions.