Distributed Optimization in Wireless Networks Using Broadcast Advantage

Abstract

In this paper, we study cross-layer design for multicasting in wireless networks by exploiting broadcast advantage. With network coding, a distributed subgradient algorithm for joint congestion control, session scheduling, and network coding is proposed, which requires a centralized scheduling algorithm in general. Under the primary interference model, we find that any valid link schedule corresponds to a hypergraph matching. To solve the scheduling problem distributedly, local greedy, randomized, and hybrid algorithms are proposed. We also modify the randomized algorithm into a constant-time algorithm. With random network coding, we obtain a fully distributed cross-layer design. Our experimental results show promising throughput gain by using our proposed framework, in some cases with less complexity than cross-layer design without broadcast advantage. Lastly, we also extend our framework to the case without network coding.