Task-oriented Distributed Information Fusion in Wireless Sensor Networks

A wireless sensor network may form a part of the base for cloud computing through Internet. We present a framework of task-oriented distributed data fusion, and investigate the assignments of heterogeneous sensors on nodes in the network for a specified task, assuming the network has been self-organized to be a temporally static hierarchy with fixed coordinators. The merits of the distributed data fusion are evaluated through a case study on a classic database from five aspects, quality of estimates, computing scalability, realtime performance, data flow, and energy consumption.