Research On Scheduling Mechanism For IEEE802.16Mesh Networks

IEEE802.16mesh network is a promising access technique for wirelss MAN,which is of broad application prospects with the advantages of low cost, widecoverage, easy to expand and high transmission rate. The scheduling mechanism ofmesh networks has a direct, significant impact on network performance indicators ofthe channel utilization, QoS and fairness.In this thesis, we make a deeply study on scheduling mechanism of IEEE802.16mesh network for network performance, and proposed improved methods forcentralized scheduling and distributed scheduling respectively.For centralized scheduling, in the scheduling fairness prospects, we proposed aHRF algorithm, which is based on the ratio of waiting time to the mean number ofrequest slots. In HRF, not only the nodes with numbers of lower traffic demand can bescheduled early to promote the network throughput, the nodes with higher trafficdemand also be considered to be scheduled in time for their service demand.For distributed scheduling, in the prospects of congestion control, we proposed adistributed scheduling algorithm based on Buffer Utility (BU), named as BU-CDS. InBU-CDS, the BU of current node and that of other competing nodes get via controlmessage broadcasting will be sorted, and the node with higher BU will be scheduledearlier, from which congestion can be avoided and end-to-end delay decreased.With the analysis and research on BU-CDS, we proposed a distributed schedulingalgorithm, named as RBU-CDS, based on the estimation of traffic load. In RBU-CDS,we note the level of traffic load of node as buffer utility. For the BU of nodecontinuously changing with time and node can not update it in time, we make BUestimation to replace collecting BUs directly. A model of game with incompleteinformation is built for estimation, in which the competitive ranking is estimateddepend on the BU of current node and joint cumulative distribution of competingnodes. Base on the estimated competitive ranking, the algorithm can make node withhigh traffic load be scheduled early.