Energy Balanced Target Tracking In Wireless Sensor Networks

Wireless sensor network (WSN) consists of a large number of low cost sensor nodesinterconnected via wireless channels to form an ad hoc wireless network. The WSN tech-nology promises high resolution spatial and temporal monitoring of expansive sensingfeld over extended duration, and hence is an enabling technology of modern cyber physi-cal systems. WSN has found numerous important applications including target tracking,infrastructure monitoring, habitat sensing, and battlefeld surveillance. Due to limitedon-board battery energy reserve, computation, processing and communication capabili-ties, WSNs must rely on collaborative signal and information processing to dynamicallymanage sensor resources and efectively process distributed information in order to pro-long the operating life-span and to reduce maintenance cost.This PH.D dissertation is based on the target tracking application in WSNs, focuseson the target tracking signal processing and nodes resource scheduling.1) For target tracking with distance-dependent measurement noise in wireless sensornetworks, a distance-dependent measurement error of sensors is modeled as a multiplica-tive noise. Then, a novel bayesian target tracking algorithm is proposed. To deal withtracking in directional sensors of a special angle of sensing range, like cameras/videosensors, infrared sensors, ultrasonic sensors, based on the original measurements anddirectional sensing range information of directional sensors that detect the target, a dis-tributed target tracking algorithm in directional sensor networks is proposed.2) As the energy imbalance of wireless sensor networks has an impact on the networklifetime, a novel energy-balanced optimal distributed clustering mechanism is proposedby introducing a new energy-balanced metric based on the standard deviation of residualenergy of nodes. Then, it is transformed into a multi-objective constrained optimizationproblem, and an adaptive genetic algorithm with Elitist mechanism is employed to resolvethis problem.3) The nodes scheduling in WSNs is a NP-Hard problem, based on a sensor-to-targetdistance estimation observation UKF tracking algorithm, a novel task-specifc defnitionof the network lifespan is given, and the theoretical connection between the energy balancemetric with WSN lifespan is established. Then, a novel, polynomial time heuristic energy balanced nodes scheduling method is proposed that are capable of delivering desiredtracking performance while signifcantly extending the WSNs lifespan. To deal with targettracking with distance-dependent measurement noise in WSNs, some similar heuristicenergy balanced nodes scheduling algorithms are presented.4) For nodes scheduling in multi-target tracking of WSNs, a three point locationis adopted, and a location accuracy model based on area sum method is constructed.Combining the energy consumption model and the location accuracy model, the compre-hensive performance index function was created to guide nodes scheduling, and geneticalgorithm was used to implement the selection. Finally, the outlook about complex targetmotions and observation models as well as its corresponding bayesian statistical estima-tions is reviewed.