Unknown source localization using RSS in open areas in the presence of ground reflections

This work explores the localization of unknown RF transmitters in open outdoor areas using Received-Signal-Strength (RSS) measurements. The positioning accuracy in RSS is typically limited by the behavior of the path-loss environment. The motivation for this work is the localization of RF interference in the GPS L1 band but as it is not legal to broadcast in that band the scenario is explored with real experiments using Wi-Fi devices. The Received Signal Strength Indicator (RSSI) output of a commercially available Wi-Fi chipset was characterized and found to be suitable for measuring path-loss by using passive monitoring of an access point and by applying a spline fit to the histogram of the incoming RSSI measurements. A sensor network of 5 sensor nodes was then setup on an oval over an area of 150m². To verify a two-ray path-loss model, RSSI was measured at a number of points at height combinations of 1.05m and 1.55m. The measured path-loss was found to be close to what was given by the two-ray path-loss model, with significant dips in the path-loss visible in the 1.55m height data. The position was then solved using the RSSI values at each test point using the log-normal path-loss model. Increasing the sensor and transmitter height was found to significantly increase the positioning error. For the data recorded for nodes and transmitter at a height of 1.05m, the Root-Mean-Square-Error (RMSE) of the test points was 10m, and for the 1.55m case the RMSE was equal to 34m. With this model verified, Monte-Carlo simulations were used to evaluate the positioning performance that could be expected in the GPS L1 band. It was observed that the positioning performance was poor if the transmitter was located too close or too far from the sensor nodes, depending on the heights of the transmitter and receivers and the sensor network topology. Hence, for robust localization to be achieved, another alternative approach to solving the positioning equations that takes in- o account the variations in the path-loss is required.