Event-based k-nearest neighbors query processing over distributed sensory data using fuzzy sets

Abstract

K-nearest neighbor (kNN) query is an effective way to extract information of interest from distributed sensing devices. Most of the existing kNN query processing approaches rely on using raw sensor readings, which is costly in terms of communication and time overhead. This paper investigates the event-based kNN query problem in distributed sensor systems and proposes a novel e-kNN query scheme using fuzzy sets. Our key technique is that linguistic e-kNN event information instead of raw sensory data is used for e-kNN information storage and in-networks kNN query processing, which is very beneficial to energy efficiency. In addition, event confidence-based grid storage method and e-kNN query processing algorithm are devised for e-kNN information storage and retrieval, respectively. Experimental results based on real-life data set further show that our e-kNN scheme outperforms the conventional methods in terms of communication cost and response time with accuracy guarantee.