Research On DCS In Wireless Sensor Networks

Wireless Sensor Network(WSN)composes of a large number of sensor nodes in monitoring area.Nodes can collect the target information of monitoring area,and will transmit the collected data by Sink node to the end of processing.If the data transmitted directly with a large number of redundant information,then it would consume more energy,an additional broadband resources,reduce the life of the Wireless sensor network.Therefore,we want to compress data in a wireless sensor network.Therefore,we want to compress data in a wireless sensor network.Wireless sensor nodes are generally tiny and cheap,the data compression is to consider the node of computing power.The characteristics of the wireless sensor network are described,and the previous data compression methods are introduced in wireless sensor networks.At the same time,a new data compression theory,which is called Compressed Sensing(CS).Compressed sensing collection a small amount of the compressed data,and transmit the compressed data to the end of processing and reconstruct it to get the raw data.Due to the compression perception has low computing complexity and the compression effect is good,it is very suitable for wireless sensor networks.According to the data correlation of different nodes,combined Joint Sparse Model and Compression Sensing of Distributed Compressed Sensing(DCS).Distributed Compressed Sensing is introduced in detail,the basic theory of compression perception are described,and the mathematical model of compressed sensing,signal sparse transformation,selection of the observation matrix and signal reconstruction are detailed analyzed.This thesis mainly researches on the Compressed Sensing in Wireless Sensor Network: the main contributions are as follows:This paper proposes an improved gradient pursuit algorithm.The algorithm on the basis of the gradient tracking algorithm proposed to the step to improve,reduce the path of the sawtooth contraction in the gradient projection method.Simulation results show that the improved gradient pursuit algorithm meet to the speed and precision of reconstruction,and the algorithm is better than existing algorithms.This paper proposes a dynamic self-organizing clusters algorithm.The algorithm achieves clustering after get the data compress by compressed sensing,so each cluster has strong correlation between internal nodes.This paper proposes the model of distributed compressed sensing algorithm based on joint sparse model JSM-1 in wireless sensor network.The algorithm is based on the concept of distributed in the wireless sensor network,joint data after cluster data processing.In practice,there are a large number of sensor nodes in wireless sensor network,using distributed compressed sensing algorithm not only can eliminate the redundant information of a single sensor node,but also can eliminate the redundant information between nodes.