Research On P2P Traffic Optimization Technologies In WLAN

This paper focuses on the research of P2P traffic optimization technologies inWLAN, including the following aspects:1) According to the analysis of the P2P trafficproblems and the P2P traffic optimization demands in WLAN, this paper designs thearchitecture of P2P traffic optimization system in WLAN;2) Based on the research ofthe traditional P2P cache technology and peer selection algorithm, this paper proposesbidirectional P2P cache and load-balancing peer selection algorithms for WLANrespectively.On the one hand, due to the characteristics of the DCF model in WLAN and P2Pusers’ requirements of both downloading and uploading, the increasing P2P trafficfrequently causes congestion in WLAN. This not only causes the problem on the P2Papplications’ own delivery quality, but also degrades the performance of the otherInternet applications in WLAN. However, traditional P2P cache can only alleviate thecongestion at the gateway but does nothing to relieve the congestion at each AP’s airinterface. To solve this problem, this paper proposes a novel solution calledbidirectional P2P cache in WLAN. A bidirectional P2P cache can be deployed at theAccess Controller of a WLAN. Most importantly, with reverse caching, the bidirectionalcache can release a considerable part of the radio uplink bandwidth resources at each APand thus alleviate the congestion caused by P2P traffic in WLAN effectively; Moreover,with forward caching, the bidirectional cache can still act as traditional P2P cache toreduce the cross-domain traffic. Simulation results show that, compared with traditionalP2P cache, bidirectional P2P cache can shorten the file transfer time of P2P applicationand improve the throughput of other Internet applications significantly. This indicatesthat bidirectional P2P cache can alleviate the congestion caused by P2P traffic inWLAN effectively, and thus optimize the overall network traffic in WLAN.On the other hand, due to the features of WLAN architecture, the famous idea of“traffic localization” is no longer suitable for wireless P2P system, this paper proposestwo novel peer selection algorithms named DB-LB-B and DB-LB-N, which can bedenoted as DB-LB algorithms as a whole. The two algorithms both based on loadbalancing policy, are designed for two different cases of the P2P application’s demandrespectively. The main idea of the algorithms is: First, adopt the load balancing policy toensure the traffic load balance on all APs; Second, considering the bottleneck of file transfer speed is usually the downlink of the receiver rather than the uplink ofbandwidth of the senders, take the requesting peer’s downlink bandwidth as the target ofthe sum of the selected peers’ uplink bandwidth; In addition, provide a Fuzzy CognitiveMap for the estimation of peers’ service ability in wireless P2P system, it cancomprehensively consider multiple related factors in wireless network environment.Simulation results show that, compared with traditional SA algorithm (always choosingpeers with the highest service ability) and RS algorithm (choosing peers randomly),DB-LB algorithms can achieve much better load balance on APs while ensuringfavorable file transfer speed. This indicates that the load-balancing peer selectionalgorithms can optimize the distribution of P2P traffic in WLAN.Finally, to validate the effect of the proposed P2P traffic optimizations in WLAN,this paper evaluates the performance of the proposed optimizations on a wireless P2Psystem simulation platform built upon OMNeT++.