LOCAL AND GLOBAL KNOWLEDGE TO IMPROVE THE QUALITY OF SENSED DATA

Sensor networks are driven by the activities of their deployed environment and they have the potential to use data that has previously been sensed in order to classify current sensed data. In this paper, we propose the Knowledge-Based Hierarchical Architecture for Sensing (K-HAS), an architecture for Wireless Sensor Networks (WSNs) that uses different tiers within a network to classify sensed data. K-HAS uses three tiers for in-network classi cation: the lower tier actively senses the data and packages it with relevant metadata, the middle tier processes the data using a knowledge base of previously classi ed sensed data and the the upper tier provides storage for all data, a global overview of the network and allows users to access, and modify classi cations in order to improve future classi cations. Initial experiments on the performance of the individual components of K-HAS have proven successful and a prototype network is planned for deployment in the Kinabatangan Wildlife Sanctuary, Malaysia.