A wireless sensor network consists of both static and mobile sensor nodes which constantly changes their position within the network and sends signal to the stationary nodes. In this type of network estimation of the nodes location becomes essential for providing a better Quality-of-Service (QoS) in the network. An optimal routing mechanism requires the location of nodes to be known to establish a reliable routing path between the mobile and static nodes. This paper explores a node localization mechanism based on the signal strength of packets received in the stationary nodes. The Received Signal Strength Indicator (RSSI) estimation is based on the okumura-hata model the location of the mobile nodes is approximated using a Support Vector Regression algorithm. The sequence of estimated RSSI values are transformed in to randomized feature space of low dimension which reduces the computational efforts required to train a regression model with a non-linear input data. The low dimensional randomized features are then used to train a Support Vector regression model with linear kernel and then Root Mean Square Error was analyzed to measure the performance of the model.
Volume 11 | 11-Special Issue