The Research Of Group Mobility Model And Routing Strategy Based On Spectral Clustering Algorithm In Opportunistic Networks

Communication between nodes has great influence on the their mobility in the opportunity network,which is frequently confronted with intermittent interruption,owing to the unstable topology network and its dynamic change.The research on group mobility in the opportunity network concerns what nodes' mobility may bring about,in contrast to the preciously-proposed group mobility approaches that are unilateral in the thoughts of the constitution of groups by nodes,resulting in,on one hand,the distortion of reality to reflect group behavior characteristics that are revealed in the situations of their application;or on the other hand,the impossibility of dynamic clustering of nodes in accordance with tasks given in any already-arranged time or place,and the ignorance of group behavior characteristics in the process of node's mobility,resulting in irrational operation.For the strategies of designing routing,the limitation of traditional approaches of community division lies in that they disregard the effects of nodes' mobility on community,the approaches of community division is static,distribution of nodes on the map is restricted and accuracy of the outcome of community division cannot be elevated efficiently.To address the problems above,the main innovations of this paper are as follows:(1)Based on the Spectral Clustering algorithm,this paper proposes the group mobility model(SCM),which takes the advantage of node's own properties,such as its spatial position,to set up node's affinity matrix,extract main characteristics from arbitrarily distributed nodes with the aid of spectral clustering algorithm,and regarding the needs of a task assigned cluster nodes dynamically in a fixed time.The model depicts truly and effectively the group characteristics of node's mobility,accomplish the conducts of group mobility and clustering well.The group mobility model based on SCM clustering algorithm following Epidemic routing protocol can be found through a series of experiments.The model that contains flexible parameters of its own and has its nodes' clustering performance well is feasible and practical.(2)Besides,given the social nature of node's mobility,the paper proposes routing strategies on the basis of the Spectral Clustering CommunityDivision(SCCD),urging to extract the information from interconnected nodes dynamically,set up node's affinity matrix,and attain the goal of community division by virtue of clustering analysis of social relationship to the affinity matrix with the application of spectral clustering algorithm,in the process of node's mobility simulation.Judging from the outcome of community division,the paper tends to conceive a message forwarding strategy that is accessible to the centrality of probability,and optimize it by using utility functions.The outcome of simulation shows that the spectral clustering community division algorithm proposed in this paper,comparing to the traditional Epidemic and Prophet,which is more efficient in transmission,possesses less expensive protocol overhead and average hop,has improved overall comprehensive performance of the network.