The wireless sensor network(WSN)is the interdisciplinary of the embedded technology,microelectronics,sensor technology,distributed information processing technology and communication network.WSN has a wide application in environmental supervision,healthcare,national defense and transportation.The use of wireless sensor network normally needs to know the specific location of the sensor nodes.The monitory information by a node is often meaningless without its knowledge of position.So,the research of node localization in wireless sensor network is of great value and significance.First,according to the signal similarity,a localization algorithm is provided.In the localization process,we first estimate the relative distance between neighboring nodes through comparing the similarity of received signal strength and then calculate the relative distance for non-neighboring nodes with the shortest path algorithm.After that,the nodes' relative relation map of the whole network can be obtained.Given at least three anchors,the physical locations of nodes can be finally determined based on the multi-dimensional scaling(MDS)technology.Second,according to the structural similarity and signal similarity,a localization algorithm is provided.This paper first draws a net topology graph throw the connectivity of nodes in the whole network.The value of signal similarity between two nodes is calculated as the weight of the topology graph edges.Then,we calculate the value of structural similarity between any two nodes as relative distance through the SimRank algorithm based on weight.Given at least three anchors,the physical locations of nodes can be finally determined based on the multi-dimensional scaling(MDS)technology.Finally,the verification experiment has been conducted by two types of experiments: a regular network and a non-regular network.Results show that the design of this paper has a good localization accuracy. |