Analysis of RSSI Feasibility for Sensor Positioning in Exterior Environment

Mobility of nodes in Wireless Sensor Networks (WSNs) brings formidable challenges to protocol design. A mobility estimation algorithm is the prerequisite for evaluating link quality, localizing nodes and excogitating a signal threshold to trigger possible handoff. The radio Received Signal Strength Indicator (RSSI) integrated in sensors has been widely used due to its low economy cost and moderate energy consumption. The distance of separation between adjacent nodes can be estimated by reading RSSI when a good portion of electromagnetic wave propagates in a line-of-sight link. However, the measurement results of RSSI fluctuate heavily because of fading signal and disturbing background noise. This paper investigates the reliability of RSSI for exterior sensor positioning. To display the one-to-one mapping between RSSI and distance, a series of static experiments are conducted and a reference curve is established. To mitigate the fluctuation of raw RSSI samples, a set of mobile experiments are carried out and five filtering methods are employed. The mitigation effects are evaluated by the Root Mean Square Error (RMSE) values. Though the overall optimal RMSE achieves 0.84, which is significantly lower than that of the raw samples, it is still possible that one RSSI corresponds to two or more distances, and the maximum difference between them can reach 2.97 meters. Because this error is intolerable for many applications, it is not authentic to gauge the distance between mobile nodes only based on RSSI in exterior environment.