Cooperative Scheduling And Distributed Information Processing In Wireless Sensor Networks

Wireless sensor networks (WSNs) are popular research fields in the domestic andinternational area, currently. WSNs involve multiple kind of technologies, such as sen-sor technology, macro-electromechanical technology, embedded computing technology,wireless network technology, wireless communication technology, and distributed infor-mation processing. WSNs are involved with cross-cutting multi-disciplinary and high in-tegrated technology. The main advantages of WSNs include low cost, rapid deployment,self-organization, and fault tolerance. Thus, it can be widely used in target tracking,environmental monitoring, medical care health, national security, tra?c management,manufacturing, anti-terrorism and other disaster areas. Information industry technol-ogy development of”the Eleventh Five-year Plan”and 2020 middle-term and long-termlayout confirm WSNs as a key technology that requires development in the future 15years.This PH.D dissertation is based on target tracking in WSNs. We mainly studyresource management and distributed cooperative information processing. So far asresource management is concerned, the specific research contexts include sensor nodescheduling, limited sensor node energy-saving, energy balance of the whole networks, andcooperative scheduling of multiple sensor nodes etc. About distributed information pro-cessing, the research is focus on filters with measurement noise and packet loss, and howto design filter with packet losses. For multiple measurements, we research how to fusethese measurements to estimate the target state. Specific contexts are as follows:1) For moving target tracking in WSNs, the dynamic scheduling idea is presented,and then is applied to the dynamic-group scheduling scheme (DGSS). The scheme notonly saves energy consumption, but also greatly increases real-time performance. Basedon DGSS, it is extended to a multiple-sensor cooperative scheduling. An energy balancemodel is constructed and a multiple-sensor cooperative scheduling algorithm is proposedbased on the energy balance model. Simulations show the presented algorithm not onlybalances energy distribution to prolong the lifetime of the whole networks, but also in-creases estimation accuracy and reliance.2)Sensor nodes have small size, limited battery energy, limited computing and com-munication bandwidth in WSNs, and moreover, sensor nodes are deployed in complexenvironment, where there are high measurement noise, packet losses and low reliance, soit is very important to design filters with packet losses. Firstly, stochastic process andinnovation analysis methods are utilized, and a kind of filters with single packet loss for linear time-varying systems is designed, and then such filters are extended from linearsystems to nonlinear systems. Secondly, the kind of filters are generalized to filters withmultiple packet losses in stochastic systems. Finally, the filters with multiple packetlosses are universalized to more generalization form, that is the minimum variance filterswith finite consecutive packet losses. The kind of filters is extended to nonlinear systems,and nonlinear minimum variance filters are designed and successfully applied to WSNsto estimate the state of moving target.3)Due to the limited resources for computation, communication, and processing,collaborative signal and information processing (CSIP) for sensor nodes is needed. Com-bined consensus strategy with Kalman filtering, distributed optimal Kalman-consensusfilters (DOKF) with multiple measurements are designed. Moreover, distributed opti-mal Kalman gain and consensus gain are derived respectively. Based on DOKF, thedistributed Kalman gain is regarded as a special form of distributed consensus gain. Weunify them into one form, and derive a kind of distributed optimal consensus filters withthe uniform form. Simulations illustrate the filters have the superior performance, suchas low estimation error and high reliability.