Design Of Opportunistic Routing Protocol Based On Stream Network Coding

Wireless Mesh networks often experience severe performance degradations when legacy routing algorithms are deployed. Opportunistic routing significantly increases unicast throughput of wireless mesh networks by effectively utilizing the wireless broadcast medium. With network coding, opportunistic routing can be implemented in a simple and practical way without resorting to complicate scheduling. Stream network coding is a method of acknowledging a coded packet before decoding in network coding-based opportunistic routing. Therefore, the implements of stream network coding have a high research value in the design of opportunistic routing protocols.Firstly, four key factors are summarized for design of opportunistic routing protocols. And we divide opportunistic routing solutions into two categories:scheduled approaches and not-scheduled approaches. Not-scheduled approaches are the trends of design opportunistic-based protocols, and we focus on analyzing network coding-based protocols. Then we summarize, analyze and compare representative opportunistic protocols, and point out the common challenges of them.After reviewing the existing protocols, we analyze the performance of classical protocol, MORE, and find that when the end-to-end ACK is send to the source node, the packets transmitted in the network is useless. It is a waste of valuable network resource. In order to solving the problem, we implement stream network coding into opportunistic routing design, and present Slide++, a coded feedback-based opportunistic routing protocol with stream network coding. In Slide++, a sliding window mechanism is proposed for source node to control the storage costs and computing complexity of all nodes in the network. Then, we find that when the packet loss rate is higher than the measurement, the throughput of MORE has a significant degradation. We improve the cumulative coded feedback for opportunistic routing with stream network coding, in order to making accurate and efficient transmission strategy for nodes based on actual reception of packets. To further limit the cost of storage and computing, a cost control mechanism is provided for intermediate and destination nodes. Finally, we evaluate the accuracy and effectiveness of Slide++using Nsclick simulator. Comparing with MORE, Slide++has higher throughputs in different network topologies. When the link quality of the networks degrades, Slide++improves throughput by up to 3.5x over MORE. Furthermore, we analyze and summarize how the size of sliding window affects the throughput of Slide++by setting different values of sliding window in simulations. And we get an empirical value of sliding window based on simulation results.