Research On Energy Efficient Compression Algorithm And Evaluation For Wireless Sensor Networks

As one of the four global high-tech industries, Wireless Sensor Networks (WSN) isdriving the rapid progress of technology and society. The great scientific value of WSNhas caused great concern in various fields in the world. As a new generation of wirelessnetworks, a considerable number of key technologies in Wireless Sensor Networks,including data compression technology, are to be studied. Data compression technologyhas become one of the focus research fields for its huge ability to effectively save theenergy consumption of network traffic and to extend the network life cycle.Realizing high-precision and low-power data collection in the ability constrainedWireless Sensor Networks, where nodes are densely distributed, is the main aim ofdata compression algorithms. Most of the existing WSN data compression algorithmsfocus on how to eliminate the correlations, which are also data redundancies, in originalmonitoring data, thereby reducing the amount of transmission data to achieve low powerconsumption. However, because of application characteristic of WSN, the type ofcorrelations are different in varies algorithms, thus limiting the ability of data redundancyremoving, and wasting the power of nodes. Moreover, rare research considers thecompression evaluation to evaluate the existing compression algorithms using the samestandard. Therefore, aiming at high energy efficiency, this paper makes great efforts todesign data compression algorithm and compression evaluation for WSN.In this paper, the research background and research status of Wireless SensorNetworks were firstly analyzed, as well as the purpose and significance of compressionalgorithm in WSN. Then the important aspects of WSN were comprehensively researched,and on base of that, the paper proposed three key points that the researchers should payattention to when studying the compression algorithm in WSN, such as compressionspecifications and requirements. After that, thorough classification and detailed analysisfor each kind of compression algorithm were researched. Finally, the compressionalgorithm named Energy Efficient Compression Algorithm Based on Three-dimensionalCorrelation in WSN (EECTC) was proposed, which was also the main work of this paper.EECTC algorithm classified data correlations into three types: multi-property correlation,temporal correlation and spatial correlation. Based on the classification, the nodes inclusters and cluster head nodes perform different compression algorithms to share energy consumption. In the detail description of the algorithm, grouping algorithm within thecluster head nodes was first designed by two defined parameters (Relevant Times andRelevant Sum), to remove multi-property correlation. Then, in order to eliminate thetemporal correlation of data, nodes in the cluster performed adaptive piecewise polynomialfitting algorithm, making the data sequence adaptively divided into several segmentsaccording to the error threshold and respectively compressed using polynomial fittingmethod. Finally, the cluster head nodes further integrated the compressed data to eliminatespace relevance. In order to verify the algorithm, simulation tests and algorithm analysiswere also included in the paper. Simulation results showed that the algorithm was ableto effectively reduce the amount of communication data in the network while compressionaccuracy requirement was met. Simulation also showed that EECTC algorithm could saveenergy consumption to a satisfying level. In the end, a new evaluation method namedCompression Energy Ratio(CER) was proposed, which established the relationshipbetween the energy consumption in nodes and commonly used evaluation parameters. Asimplified performance assessment method on the basis of one-order of wireless energyconsumption model was also proposed. Then at last, the analysis and simulation resultsproved the validity of the evaluation model, and once again proved the good performanceof the proposed EECTC algorithm.