Research On Compressed Sensing Algorithm For Airport Noise In Wireless Sensor Network

The traditional monitoring mode,with high cost,strict requirements for environment and few monitoring sensor,is unable to get comprehensive accurate measurement of the airport noise.Wireless Sensor Network(WSN) is a multi-hop,self-organizing network,Composed by plenty of cheap sensor nodes deployed in the monitoring area, the object can be monitored all-round and all-weather.WSN is a network of limited energy,compared with traditional networks. It is the greatest concern how to reduce the consumption of the network as much as possible while ensure the accuracy of the data. Data fusion technology becomes the research focus in recent years as it can reduce energy consumption and prolong the survival time of WSN.Compressed Sensing(CS),as a new kind of data fusion technology,breaks out the limitations of traditional signal sampling, provides a method to reconstruct the original signal from less sampling data in high probability,and it can reduce the redundant information in the sensing data greatly, thereby reducing the amount of data transmission network is widely used in many fields.Firstly,we make a comparative analysis of several classic greedy algorithms: Orthogonal Matching Pursuit(MP), Subspace Pursuit(SP), Sparsity Adaptive Matching Pursuit(SAMP),Forward Backward Pursuit(FBP).Furthermore,the Linear Variable Step Forward-backward Pursuit Algorithm(LvsFBP) is proposed based on FBP,LvsFBP adjusts the iteration step size adaptively which can reducing operational costs while enhancing algorithm reconstruction accuracy.Then,Base-Station Controlled Classification Clustering And Compressed Sensing(BCCACS) is proposed to solve the airport noise monitoring issues. BCCACS utilizes Base-Station controlled classification clustering to splits clusters into balanced size using iterative clustering technique and combined with the airport environment and node location information,then broadcasts clustering information;each sensor compresses sampling signal according to temporal correlation and sends data to cluster head;CH utilizes spatial correlation for further compression;CH selects the optimal neighbor non-CH node for forwarding;BS reconstructs sampling signal using LvsFBP algorithm.Simulation results on NS-2 show that BCCACS can significantly improve the network performance of clustering,make all clusters have approximately the same number of nodes, reducing the amount of network data,balance the network transmission energy consumption,and enhance the network lifetime.