Localization in Wireless Sensor Networks by learning movement pattern using Hidden Markov Model

Abstract

The Sensor Network Localization problem deals with the estimating the geographical location of all nodes in Wireless Sensor Network (WSN),focusing that sensors to be equipped with GPS, but it is often too expensive to include GPS receiver in all sensor nodes. In the proposed localization method, sensor networks with non-GPS nodes derive their location from limited number of GPS nodes. The nodes are capable of measuring received signal strength (RSSI) and\or are able to do coarse (sectoring) or fine signal angle of arrival (AoA) measurements. Existing algorithms exploit the aspects of such sensory data to provide either better localization accurary or higher localization coverage. However, there is a need for a framework that could benefit from the interactions of nodes with mixed types of sensors, in certain applications RSSI measurements of single sensory capacity sensor nodes are most appropriate for better localization. Most of the sensor nodes has unique movement profile, which is possible to learn and used to predict location. In the proposed work, localization is achived by Bayes Particle Filter method and Hidden Markov Model (HMM). From the results Learning movement patterns HMM gives faster predictionand improved error estimates.