Research On Key Technologies Of QoS Based On Wireless Mesh Networks

With the highly development of telecommunication technology, People have more and more requirements to network access, including to access the network everywhere, to access the network with high bandwidth, to access the network with high QoS. Wireless Mesh Networks(WMNs), which is evolved from Mobile Ad-Hoc Networks, is definitely an effective distributed wireless access solution to match the people's high bandwidth requirement and can access the networks everywhere..Because of the high QoS requirements from multimedia services, VoIP service, Video Conference service and so on, more and more people wish to acquire such services from WMNs and expect WMNs can provide almost the same QoS ability as wired networks. Especially multimedia services, which are very sensitive to packet loss ratio, delay and delay variance, are more and more popularized. However WMNs have to face the big challenges to provide such kind of QoS ability because of dynamic topology, multi-hop and limited bandwidth resource, even it has the advantages to self-organize and make network access very easy. Some features, like shared wireless channel which are competed by all wireless nodes in the range and inaccurate link state maintained by wireless nodes, will make QoS over WMNs very difficult compared to wired networks. The right approaches to improve the QoS ability of WMNs are to investigate current QoS technologies from wired network and MANETs, and adapt those excellent solutions to WMNs and finally implement the QoS of WMNs which is better than"Best effort". This dissertation will include following researches based on current QoS solutions in wired networks and MANETs:Firstly, some multipath routing protocols in MANETs are investigated, which include their ability to improve the QoS parameters, their advantage and disadvantage for load balance. Based on these researches, a new multipath routing algorithm with load balance, named ONDMR, is introduced in the dissertation. The algorithm can calculate the correlation of multiple routes without increasing the route overhead obviously, through adding some fields in RREQ and RREP messages. After calculating the correlation of each two route pair and route hop number of each route, ONDMR will distribute the data packets over multiple routes based on the weight which is calculated from correlation and route hop number. Simulations over ns-2 show that it has better performances when nodes mobility is low, especially over high data traffic.After investigating current Ad Hoc QoS routing algorithms, a new solution of QoS routing algorithm, named QNDMR, was included in the dissertation. The algorithm will recover multipath between source node and destination node, which have enough free bandwidth not just on the route nodes, but also on the contention neighbors of those route nodes. Over the multipath routing algorithm, admission control procedures based on free bandwidth, which are invoked by receiving RREQ, RREP and Hello messages, are used by recovery of multipath routes. At the same time simulation over ns-2 shows that delay and packet delivery ration are better that AOMDV and NDMR when the entering data traffic is very high.Secondly, MAC layer always plays an important role of QoS ability in WMNs. In the dissertation, Contention-based admission control procedure for IEEE802.11e, named EDCA, is investigated, high priority services are impacted by low priority services especially with high data traffic of low priority services To resolve the problem, a new admission control procedure with multiple RTS requests for high priority services is introduced in the dissertation, the new algorithm will decrease the collision possibility of high priority services, simulation over ns-2 shows the new algorithm will increase the successful possibility to admit high priority services.At the same time, a new adaptive algorithm for congestion control, named collision avoidance BEB(CABEB), is also introduced in the dissertation. In the CABEB algorithm wireless nodes will detect the congestion based on the length of transmission queues, the queue length information, named contention index - CI, will be piggyback to transmission node on ACK frame, the transmission nodes will select different contention windows for services with different priority based on the CI; When the receiving node is in extreme congestion state, it will reply the NTS frame againt RTS. Simulation over ns-2 shows that the new algorithm will improve the QoS paramenter to services with high priority.