Adaptive Network Coding and Scheduling for Maximizing Throughput in Wireless Networks
ACM Mobicom |
Recently, network coding emerged as a promising technology that can provide signiﬁcant improvements in throughput and energy eﬃciency of wireless networks, even for unicast communication. Often, network coding schemes are designed as an autonomous layer, independent of the underlying Phy and MAC capabilities and algorithms. Consequently, these schemes are greedy, in the sense that all opportunities of broadcasting combinations of packets are exploited. We demonstrate that this greedy design principle may in fact reduce the network throughput. This begets the need for adaptive network coding schemes. We further show that designing appropriate MAC scheduling algorithms is critical for achieving the throughput gains expected from network coding. In this paper, we propose a general framework to develop optimal and adaptive joint network coding and scheduling schemes. Optimality is shown for various Phy and MAC constraints. We apply this framework to two diﬀerent network coding architectures: COPE, a scheme recently proposed in , and XOR-Sym, a new scheme we present here. XOR-Sym is designed to achieve a lower implementation complexity than that of COPE, and yet to provide similar throughput gains.