Research On Wireless Network Key Technologies Based On Link Stability And Network Coding

Wireless mesh network has attracted the interests of scholars and engineers due to its good flexibility and scalability. At the same time, as the number of network user increases sharply and demand of network service varies, more and more researchers have paid their attention to optimizing network resources and maximizing their usage. Network coding is an important breakthrough in the fields of network communication. By applying network coding, the intermediate nodes can mix the packets they received according to pre-specified rules before forwarding, thus network performance can be greatly improved. This dissertation focuses on how to find a stable path in a network with dynamic topology and how to utilize network coding to improve network performance. The main innovative results are listed as follows:Firstly, link stability based routing scheme LSB_AODV is proposed after summarizing existing stability based routing scheme. LSB_AODV selects a stable path while minimizing hop counts instead of selecting the shortest path used by AODV. Link stability factor (LSF) and path stability factor (PSF) are used in route selection. The node's transmission range is divided into stable zone and caution zone. Rerouting is initiated when the distance between two neighboring communicating nodes exceeds the stable zone, before the old path actually breaks, thus realizing soft handoff and reducing end to end delay. The proposed LSB_AODV routing scheme is evaluated by simulation and results show that an improved packet delivery ratio and reduced ETE delay can be obtained.Secondly, the first practical wireless network coding system COPE can not actively detect a route with more coding opportunities and limit the coding structure within two-hop regions. Thus, an on-demand coding-aware routing scheme (OCAR) for wireless mesh networks is proposed to overcome the limitations specified above by actively detecting a route with more coding opportunities along the entire route rather than within two-hop regions. Utilizing more coding opportunities tends to route multiple flows "close to each other" while avoiding interference requires routing multiple flows "away from each other" OCAR achieves a tradeoff by adopting RCAIA as routing metric in route discovery, which is not only coding-aware but also considers both inter and intra flow interference. Simulation results show that, compared with AODV and AODV+COPE, OCAR can find more coding opportunities, thus effectively increase network throughput, reduce end to end delay and alleviate network congestion.Thirdly, a successful delivery probability based coding-aware opportunistic routing mechanism is proposed in which successful delivery probability (SDP) is used as routing metric instead of traditional routing metrics such as hops and expected transmission count (ETX). Forwarder set is adopted and all the nodes in the forwarder set can perform coding and forward packets according to their priority in a distributed manner, thus the problems such as inter-flow interference and overloaded intermediate nodes caused by traditional coding-aware routing protocol can be overcome. Simulation results show that the proposed mechanism can obtain more coding opportunities at the cost of increasing a small amount of duplicates, thus effectively promote network performance.Finally, a priority based network coding (PBNC) scheme is proposed given the stochastic nature of packet arrival process and various traffic demands on bandwidth, end to end delay and etc. In PBNC, whether a packet is transmitted with or without network coding is determined by the buffer's queue state of the relay node, thus the limitation of large delay and packet loss rate with traditional network coding (TNC) can be overcome. In addition, PBNC offers higher forwarding probability to flows with higher priorities, which can satisfy different priority demands of different flows. Markov Chain is used to model the system and parameters such as delay and packet loss rate are derived from the model. Simulation results validate our theoretical analysis and demonstrate the superiority of PBNC over traditional network coding.