Distributed Sensor Scheduling For Target Tracking

Wireless Sensor Networks (WSNs) have been considered as promising technique for area surveillance applications, and mobile target tracking is essential for these applica-tions. In non-cooperative target tracking systems, which represents the general scenario, sensor nodes need to detect and sense the target actively by emitting energy and mea-suring the feedback. Then the inter-sensor interference problem is frequently presented if nearby active sensors work simultaneously. Thus sensor scheduling is essential to avoid the interference and guarantee the tracking performance.Though there are a great deal of solid research on sensor scheduling in WSNs, most of the research studied the tradeoff between tracking performance and energy consump-tion and can not be applied directly to the interference problem for active sensors. Mean-while, due to the nature of WSNs, sensor scheduling has to be performed in distributed and light weighted ways. Thus, this thesis has the following work to overcome these chal-lenges:This thesis proposes a dynamic distributed sensor scheduling (DSS) algorithm, where the tasking node is elected spontaneously from the nodes near the target via random com-petition based on CSMA. The channel will be released immediately when sensing task is completed. Analysis reveals that DSS can avoid most of the interference and both simu-lation results and testbed experiment demonstrate the effectiveness of DSS for large scale sensor networks in terms of system scalability and high tracking performance.This thesis also studies the static sensor scheduling. With stationary network topol-ogy, TDMA strategy is adopted for the consideration of collision-free and energy ef-ficiency. Then by graph theory, the scheduling problem is transformed into a coloring problem which aims at minimizing the number of used colors. Since the original problem has been proved to be NP-hard, a distributed saturation degree based algorithm (DSDA) is proposed. DSDA can be implemented locally by each node with information collected from its neighbors. Then the correctness of DSDA is proven as well as analytical results for the complexity and energy consumption of the algorithm are provided. Extensive sim-ulations also demonstrate the effectiveness of DSDA.At last, combining the dynamic sensor scheduling and the static one, this thesis de-signs a hybrid distributed sensor scheduling (DHSS) algorithm. DHSS is based on DSDA and uses the time slot assigned by DSDA. When empty slot appears, the random competi-tion in DSS is adopted to reuse the slot. Thus DHSS can achieve better tracking accuracy than DSS and DSDA with appropriate energy consumption, which is larger than DSDA but smaller than DSS. Simulation results also validate that DHSS obtains this design goal.