Resource-aware Distributed Online Data Mining for Wireless Sensor Networks

Online data mining in wireless sensor networks is concerned with the problem of extracting knowledge from a large continuous amount of data streams with an in-network processing mode. Unlike other types of networks, the limited computational resources require the mining algorithms to be highly efficient and compact. We propose a distributed resource-aware online data mining framework for wireless sensor networks which can be used to enable existing mining techniques to be applied to sensor network environments. We have applied the framework to develop and implement a distributed resource adaptive online clustering algorithm on the novel Sun Microsystem Small Programmable Object Technology Sun SPOT platform. We have evaluated the performance of the algorithm on the actual sensor nodes. Experimental results show that the clustering algorithm can improve significantly in resource utilization while maintaining acceptable accuracy level.