Research On The Performance Measurement Method Of Virtual Access Points In Software Defined Wireless Networks

Software defined wireless network(SDWN) is referred as one of prospective key techniques for the fifth generation mobile communication network(5G), which has been changing traditional network architectures seal and rigid features. What's more, SDWN can satisfy the requirement of high bandwidth for mobile users' access and optimize the management of network resource. In SDWN, ISP can establish independent wireless network by using the same AP equipment on different virtual AP to realize wireless terminal access. However, there are so many virtual APs in large-scale SDWN network,full-mesh direct measurement of VAPs' performance is challenging, inefficient or even unfeasible to monitor each and every VAP due to limited storage resources and processing power.Aiming at the above problems, this thesis researches on the performance measurement method of VAP in SDWN. In order to realize a controllable measurement for all the VAPs, this paper respectively based on libpcap open source library, Netlink mechanisms and OpenFlow protocol proposes VAP's three performance indicators measuring method, like signal strength, the number of associated users and traffic(SUT). In order to solve the problem of recovering from a small number of real time measurement data to obtain the whole network data set, a method of adaptive and sequential sampling based on matrix completion(MC-ASS) is proposed, based on few VAPs' direct measurements, MC-ASS algorithm uses matrix completion recovery method for estimating the other virtual AP corresponding performance value. In order to improve the prediction accuracy of the algorithm, a MC-ASS optimization mechanism is proposed, this thesis clusters all virtual APs under a physical AP before measuring,and put the matrix of VAPs' users data as the training sets of VAPs' traffic, optimizes these two parameters' sample strategy in MC-ASS algorithm. After compared with previous, result shows that the optimized sample strategy can further improve measurement accuracy while ensuring high stability.This thesis realizes the mentioned above algorithm on SDWN platform, which is built with general AP equipment and floodlight controller, and compared the test result with common Wi-Fi tester. The results indicate that the performance measurement method for VAP can realize virtual AP performance measurement of the entire networkand significantly reduce the measurement sample rate and error while ensuring stability.Based on the research on VAP performance measurement method, it can reduce AP's load and save the storage resources. What's more, the measurement data can facilitate the research on load balancing, handover, energy saving and other network optimization works.