Research Of Clustered WSN Data Collection Method Based On Adaptive Compressive Sensing

The wireless sensor network(WSN)are usually composed of a large number of energy-limited sensor nodes,which communicate in a wireless multi-hop routing mode.These decentralized nodes can monitor,perceive and collect information of various environmental objects in real time collaboratively.At present,they have been widely used in various fields.However,the amount of sensing data in the network is very large,which makes each node need to consume a lot of energy in the process of wireless transmission,but the sensor node itself usually does not have a continuous power supply.Therefore,on the premise of guaranteeing the quality of data transmission,reducing the network energy consumption by reducing the amount of data transmission is of great significance for prolonging the life cycle of the network.Compressive sensing technology breaks the limitation of the traditional Nyquist sampling law on data sampling frequency,and simultaneously collects and compresses the sensing data in WSN,so that the original sensing data can be reconstructed accurately using a small amount of data observations.However,simply applying compressive sensing to each sensor node can not effectively reduce the network data transmission.It also needs to design better data collection methods according to the network structure and the characteristics of the sensing data.The clustered WSN has the advantages of strong robustness and network load balancing,so this thesis takes this network structure as the research object.Firstly,a data collection method with adaptive sampling rate is designed according to the linearity of WSN sensing data;Secondly,in order to optimize the reconstruction algorithm and consider the unknown WSN data sparsity,a threshold-based variable step-size sparsity adaptive matching pursuit algorithm(TVsSAMP)is proposed,which improves the clustered WSN data collection method by using adaptive compressive sensing technology.The specific work is as follows:Firstly,the characteristics of WSN are introduced,the basic theoretical framework of compressive sensing and its three key steps are briefly described,and the research status of WSN data collection methods based on compressive sensing is analyzed.Although the traditional data collection method using compressive sensing technology in sensing nodes can reduce the amount of network data transmission to a certain extent,the value of the amount of data is still large.To solve this problem,this thesis designs a more efficient network data collection method by adding adaptive compressive sensing algorithm based on the hybrid compressive sensing clustered network data collection method.Secondly,based on the clustered WSN,a network data collection method with adaptive sampling rate adjustment is proposed.The data collection method based on Clustering with Hybrid CS can effectively reduce the amount of data transmission and balance the networkload.However,the change of signal sparsity in time and space can not be taken into account in fixed sampling rate,which makes it difficult to guarantee the quality of signal reconstruction at low sampling rate,while high sampling rate will result in waste of resources.To solve this problem,a clustered network data collection method with adaptive sampling rate adjustment is proposed based on data linearity analysis.Firstly,Sink nodes analyze the linearity of reconstructed data between the current sampling time and the previous sampling time in order to grasp the trend of data change;Then,according to the analysis results,the sampling rate needed by the network at the next sampling time and the observation dimension required by cluster head nodes are calculated;Finally,the data transmission tree is used to adjust the number of cluster head nodes.The simulation results show that,compared with the network data collection method based on fixed sampling rate,this method can effectively improve the reconstruction accuracy of compressed data with a small amount of data transmission required by the dimension of feedback observations.Finally,in view of the unknown data sparsity in WSN,the Threshold-based Variable-step Sparsity Adaptive Matching Pursuit(TVsSAMP)algorithm is proposed.Combining the Backtracking-based Adaptive OMP(BAOMP)algorithm with the Sparsity Adaptive Matching Pursuit(SAMP)algorithm,which has good reconstruction performance and does not need to know the sparsity of the original signal in advance: Firstly,in the stage of atomic selection,the number of atoms with estimated support set is controlled by adaptive threshold to improve the accuracy of atomic selection.Secondly,in the iteration process,the update step of the algorithm is adjusted adaptively according to the energy difference between the residuals of each round,which solves the contradiction between the number of iterations and the reconstruction accuracy.The simulation results show that TVsSAMP algorithm can better balance the relationship between reconstruction accuracy and reconstruction time than other similar reconstruction algorithms.When the TVsSAMP algorithm is applied to the clustered WSN data collection method with adaptive sampling rate adjustment,the final reconstructed data collected by the sink nodes is more accurate.