A methodology to optimize query in wireless sensor networks using historical data

In wireless sensor networks (WSN), a query is commonly used for collecting periodical data from the objects under monitoring. Amount of sensory data drawn across WSNs by a query can significantly impact WSN's power consumption and its lifetime, since WSNs are battery operated. We present a novel methodology to construct an optimal query containing fewer sensory attributes as compared to a standard query, thereby reducing the sensory traffic in WSN. Our methodology employees a statistical technique, principal component analysis on historical traces of sensory data to automatically identify important attributes among the correlated ones. The optimal query containing reduced set of sensory attributes, guarantees at least 25% reduction in energy consumption of WSN with respect to a standard query. Furthermore, from reduced set of data reported by optimal query, the methodology synthesizes complete set of sensory data at a base station (reporting unit with surplus power supply). We validated the effectiveness of our methodology with real world sensor data. The result shows that our methodology can synthesize complete set of sensory data analogues to standard query with 93% accuracy.