RSS based Localization of Sensor Nodes by Learning Movement Model

Node Localization in Wireless Sensor Networks (WSNs) is widely used in many applications. Localization uses particle filter that provides higher network traffic due to continuous updates, which leads to high power consumption. The article presents a range-based localization for Mobile Nodes (MN) that builds up on Hidden Markov Model (HMM) algorithm. The proposed work is based on MN and the state is hidden in the Received Signal Strength (RSS) for outdoor applications. Hidden states uses explicit knowledge of the observation probability obtained from two-ray ground propagation model. HMM correlates these observations to predict the hidden states. The state transition and the observation of HMM help to estimate the most probable state sequence and the last state obtained is the predicted location. This work uses various mobility models for the movement of nodes. Varying the transmission range effectively controls the network connectivity. Results from simulation study have revealed the possible reduction of network traffic and power consumption with less estimation error. In addition, this work provides an efficient confidence interval for the estimation error. Key-Words: Estimation Error, Hidden Markov Model, Localization, Mobile Nodes, Received Signal Strength, State Estimation, Wireless Sensor Networks.