The Research On Critical Node Prediction Algorithm In Opportunistic Networks

With the fastly development of hardware and software in wirelesscommunication technology.Wireless Sensor Network(WSN) which has thecharacteristic of cheap、 low power、 self-organization is borned.It is muti-hopself-organized netowork which deploys large number of sensors to work together indetection area.Now it is more and more widely used in the field of environmentdetection、healthy care、military field and target tracking.As the mobility of the node, Wireless Sensor Network has been improved fromthe traditional stationary network to dynamic network.With the diverse the feature ofnetworks, relative problems to be considered are quite different.Energy control、network coverage,data fusion is the main study field of traditional stationarynetwork.In contrast, due to the mobility of nodes, network connectivity, forwardingmechanism, node movement model becomes the main research field of dynamicwireless sensor networks. This paper is mainly targeted at predicting the critical nodein the oppotunistic network.With the help of nodes’ mobility to make them access, oppotunistic network canbe a self-organized network without a complete path between the source node and thedestination node.The critical node in the oppotunistic network are thosewhose missing will cause the network break up. In order to make a furtherbreakthrough, this paper has reasearched almost the relative algorithms of detectingcritical node by classification and comparison.With the study of critical node detection Algorithms in wireless sensor network,a critical node of oppotunistic network prediction algorithm is proposed which basedon the historical connectivity data.Consideing the influence from the connected timeand frequency between the nodes to connectivity propability, we figure out a new wayto compute the connectivity probability.To make the study simply, time is sliced intocycles.The probability of every cycle can be computed by using the historicalconnection Information.Through analyzing the trend of the computed probability, theexpected probability can be predicted by the method of Exponential Smoothing whichis based on the trend of the historical probability.A graph which connected byprobability can be made with the predicted connectivity probability.After delete therelated node from the graph,we can get the probability of its being the critical node.To verify the reliability of the simulation scheme, large numbers of simulation experiences are launched. The experimental results show this algorithm can predictthe critical node in the oppotunistic network with a high accuracy.