Research On Task-oriented And Adaptive Management Technology For Wireless Sensor Networks

Wireless sensor networks (WSNs) will be used in a variety of applications such as environmental monitoring, military, medical, and public safety systems for their good flexibility, robustness and intelligence. Compared with traditional network systems, WSNs have some new characteristics such as tight coupling to the physical world, resource constraint, large scale and dynamic conditions. In order to coordinate the whole network and efficiently perform the sensing task by utilizing the available resources, WSNs management technologies are needed.Based on the application of battlefield surveillance systems and NSFC project of "Research on task assignment policies and implementation mechanisms in wireless sensor network", this thesis focuses on the research of WSNs management technologies. The main contributions include:? An integrated management framework was proposed for wireless sensor networks, which performs system management with closed-loop feedback mechanism, and uses policy-based management to integrate system management as well as network management.? Macro-micro management architecture was proposed for large scale heterogeneous layered WSNs. Macro-manager makes high level strategic decisions about how to optimize the available resources to achieve the mission objectives and distributes management policies to lots of distributed micro-managers, while micro-manager schedules sensor nodes to best carry out the requests from the macro-manager. Using the architecture, management complexity and communication overload are reduced.? Distributed management structure for WSNs was constructed using self-organizing clustering scheme. In large scale WSNs, clusters can be formed and micro-managers were executed in cluster heads to manage sensor nodes efficiently. For heterogeneous WSNs, the optimal number of clusters was calculated according to energy efficiency, and the performance of clustering algorithm based on Voronoi diagram was improved by limiting the flooding range of broadcast messages. For homogeneous WSNs, a single-hop clustering algorithm based on maximal independent set was proposed and a backbone connecting cluster-heads with the sink node is formed with smallest cost to exchange information among managers.? As sensor nodes usually have limited energy which mainly used for communication, a scheme which selects appropriate key sensor nodes to perform energy-conserving processing tasks was proposed to reduce communication overhead.? Optimal selection of nodes to perform energy-conserving processing tasks is a NP-Complete problem and there exist invariant and transformable processing tasks in WSNs. For invariant processing tasks, the task assignment was formulated as a quadratic 0-1 programming problem, and a layer-by-layer...