Sound ranging using multilateration and Kalman filter

Abstract: Sound ranging is a method to locate sources of sound waves using microphones or other types of receivers at known reference positions. Nowadays, these receivers can be made both very small and powerful, lasting for a long time on a single battery charge, making them suitable for longtime outdoor purposes such as detecting artillery fire or other loud sounds. In this master thesis, we present a sound ranging algorithm based on threedimensional multilateration, Kalman filtering and sound detection. Results from simulations are given. Results from field experiments are presented. Furthermore, we compare the two and and evaluate their differences. Finally, suggestions on future work are provided. The evaluation shows that while the Kalman filter and detection algorithm performs well even with high levels of measurement noise, the multilateration algorithm can provide an accurate source positioning, but it is very sensitive to errors, with performance degrading heavily due to low sampling resolution, too close node positioning, undefined sound peaks and a lack of robustness in the multilateration algorithm.