Research On Forwarding Strategy With Congestion Control In Opportunistic Networks

Thanks to the proliferation of smart phones and wearable devices,mobile social networks and pocket switched networks has become to two typical and important subclasses of opportunistic networks.In these opportunistic networks,most forwarding strategies were designed based on the social properties where messages are transferred to the node with higher value of social properties when contact occurs.Obviously,these forwarding strategies can improve the directionality of message transfer.But they can also cause an unbalanced traffic distribution.Storage congestion is very likely to happen to these nodes with highest social properties when the network traffic demands increase.This thesis discusses the related design problem of forwarding strategy and congestion control with the objective to improve the stability and robustness of opportunistic networks.Firstly,this thesis analyzes the social properties of the nodes in the opportunity networks composed of portable devices.Based on that,a new efficient method to measure the node social properties is proposed.Social centrality and social relationship strength are used in this new method.Social centrality represents the node's total connection degree with all other node in the network.Social relationship strength is used to measure the interaction strength between two nodes,and it includes social link measurement and social similarity measurement.Then,to analyze the changing tendency of network characteristic quantities with respective to different degree of network congestion,this thesis amends and improves an existing Markov chain based model of message transfer in opportunistic networks.According to the numerical results of this model,this thesis discusses the feasibility of congestion detection via these network characteristic quantities.At last,this thesis proposes a social property based forwarding algorithm with congestion control mechanism.The proposed algorithm combines the techniques of congestion avoidance,offloading and drop policy to solve the congestion problem of opportunistic networks.This thesis uses the ONEs simulator and the Infocom05 and Cambridge connection traces to evaluate the effectiveness and efficiency of the proposed algorithm.The simulation results show that the proposed algorithm can improve the message delivery rate and reduce the total dropped packets compared with the benchmark routing algorithms like epidemic routing,PROPHET and SimBet when the network load is heavy.Moreover,the proposed algorithm does not introduce significant additional delivery delay due to congestion control mechanism compared with the other three algorithms.