Congestion Control And Data Scheduling Research For Concurrent Multipat Transfer In Wireless Mesh Networks

Wireless mesh network has become the focus in the next generation network due to its advantages,such as flexible networking,wide coverage and high reliability.Concurrent Multipath Transfer(CMT)technology can effectively aggregate network bandwidth,improve resource utilization and system robustness,becoming the research emphasis of network technology in recent years.The mesh structure of wireless mesh network is able to provide multiple paths,which favors the concurrent multipath transfer of data.Therefore,the research of CMT in the wireless mesh network is of great significance to improve the network performance.Whereas,CMT could lead to fairness problem between multipath flow and single path flow when sharing bottleneck link.Moreover,CMT also produces problems such as out-of-order and cache congestion at the receiver.The problems not only reduce the throughput of the system but also affect the popularization and application of CMT in the network.Reasonable congestion control and data scheduling mechanisms can effectively address the problems,which have great research value.Therefore,this thesis goes into details of these two mechanisms,and the main research content is as follows:When packet loss occurs,existing congestion control mechanisms for concurrent multipath transfer estimate path status coarsely,and update the congestion window simply,failing to effectively meet the three goals of congestion control for CMT,that is,fairness,throughput improvement and equilibrium of congestion.Therefore,this thesis proposes a congestion control algorithm based on the level of congestion and data rate.By precisely evaluating the congestion level of the path with the data amount backlogged at the router,the algorithm can realize the fine-grained control of the congestion window;In addition,by judging the interactive relationship among multiple paths based on data rate,the algorithm achieves the joint control of congestion window decrease;Meanwhile,the increase of the congestion window is optimized based on the fairness principle and the congestion window decrease mechanism.Then,we use network simulation platform NS2 to verify the algorithm.The simulation results show that the algorithm can ensure fairness on the bottleneck link,and realize the equilibrium between multipath congestion.At the same time,compared with the existing congestion control algorithms,it can effectively improve the system throughput,and optimize the network performance.Scheduling data by predicting the order and time of packets arriving at the receiver can effectively reduce the possibility of out-of-order.However,the existing scheduling algorithms ignore the influence of the path's reliability when conduct predicting.Therefore,this thesis proposes a data scheduling algorithm based on reliability prediction.First,evaluate the quality of each path with bandwidth and delay.Second,select the effective path set for concurrent multipath transfer according to the quality of path.Finally,predict the arrival time of data by introducing path's reliability factors,that is,packet loss rate and congestion level.The path in effective path set which arrives at the receiver most quickly and has good congestion state will be selected as the path for transfer.We simulate the algorithm with NS2 platform.The simulation results show that the performance indicators of the designed algorithm such as average throughput,end-to-end delay,reorder delay and receiving cache size all outperform that of the other scheduling algorithms.The designed algorithm can reduce possibility of packet's out-of-order,improving the transmission efficiency.In the end,the thesis summarizes the full text and the future research work is prospected.