Novel RSSI evaluation models for accurate indoor localization with sensor networks

In a Wireless Sensor Network (WSN), it is often necessary to know the accurate location of one or more nodes. Various localization methods have been proposed, namely: TDOA, TOA, AOA, GPS and RSSI. All methods except RSSI require added hardware which increases the cost of the WSN. Moreover, the RSSI value of received radio signals is easily available on most sensor nodes. We propose a set of three techniques (mean+filter, mode, mode+filter) for evaluating RSSI for accurate indoor localization. The techniques are developed with the help of rigorous experimental data collected with TelosB nodes. Localization performance with all the models is compared with an existing technique, the mean technique. We assert that the mode-filter technique achieves a 64.86% less error over the mean technique in distance estimation between the transmitter and the receiver, and it has a 21.33% lower error in location estimation.