A Study On Target-perception-oriented Self-organization Of Wireless Sensor Networks

Wireless sensor networks (WSNs) are formed via self-organization by a largenumber of micro-sensor nodes with low cost, low energy consumption and thecapability of sensing, signal processing and storing, and wireless communication.WSNs have drawn significant attention from the academic circles, militarydepartments and industries for their wide potential applications. This dissertationaddresses on the target-perception-oriented self-organization of WSNs, in which thecoverage and tracking problems are emphasized, and the main contributions are asfollows:1. An index function of integrated performance including the informationacquisition and energy conservation is constructed. In addition, adynamic collaboration algorithm for self-organization (DCS) is thenproposed. The new algorithm can dynamically select leaders and theirmembers of clusters on a case-by-case basis, and the energyconsumption can be considerably reduced.2. According to the characteristics of single/multiple targets tracking inWSNs and based on the proposed DCS algorithm, a single-targettracking algorithm and a multiple-target tracking algorithm with stateaugmentation are proposed. Simulation results show that thesingle-target tracking algorithm can achieve both high-accuracytracking and low energy consumption through dynamic coordinationand information fusion of sensor nodes, and the multiple-target trackingalgorithm performance can be greatly improved by introduction ofvelocity state.3. Taking the application of the dynamic collaborative algorithm in targettracking as an example, some typical information utility functions areanalyzed and simulated for the self-organization algorithms ofperception oriented WSNs. On this basis, the algorithms and theirrobustness are discussed. Finally, the selection of information utilityfunctions for different cases are given, which can provide guidance forthe system design of WSNs.4. For the case of high false alarm and low detection probability, an emotional ant based self-organization algorithm is proposed throughoptimizing the wake-up probability by coordination of sensor nodes.The new algorithm can arouse more nodes close to the target so that ahigh detection probability of the target is preserved; while let themajority of nodes in the whole surveillance region stay asleep so as tosave energy. The effectiveness of the proposed algorithm isdemonstrated by simulations.5. An optimized probabilistic planning coverage algorithm is proposed,which can achieve the deployment of WSNs for converge of arbitrarypercentage. Two theorems are exposited respectively about therelationship between network deployment and coverage percentage andthe relationship between network connectivity and coverage, based onwhich a novel probabilistic waking mechanism is proposed for workingnodes selection. The key parameter design is also studied accordingly.Simulation results show that the proposed algorithm can effectivelyobtain the coverage of arbitrary percentage as well as a betterperformance for full coverage.6. A test system for WSNs is established. Network performance andlocalization algorithm for WSNs are tested via the RSSI-basedlocalization system. Besides, a typical sensor model and the targetdetection algorithm based on CFAR are analyzed and tested viaexperiments on acoustic sensors. The obtained results establish aconcrete foundation for further researches and applications of WSNs inthe future.