Position Estimation of Multiple Signal Sources in Short Baseline Environments Using Time Difference of Arrival and the MUltiple Signal-source Cross-correlation based Location Estimation (MUSCLE) Algorithm

Abstract

This thesis uses time difference of arrival (TDoA) measurements to obtain position solutions for multiple radio frequency (RF) signal sources in short baseline environments. Details for TDoA generation are provided along with analysis of how this measurement is challenged by both multiple signal sources and short baseline environments. The MUltiple Signal-source Cross-correlation based Location Estimation (MUSCLE) algorithm is introduced as a novel method to position multiple signal sources using TDoA. This algorithm incorporates maximum likelihood estimation (MLE), combinatorics simplifications to the solution space, and K-means machine learning to extract multiple position solutions. To test this algorithm, a simulation space is created and fully detailed. This simulation is used to test a wide range of algorithm applications without the limitations of hardware setup. Results for the MUSCLE algorithm are applied to several simulation scenarios including an ideal scenario, varied signal sampling rate scenarios, and varied receiver geometry scenarios. The ideal scenario shows that the MUSCLE algorithm can locate all signal sources in a 30x30 meter environment to less than 0.4 meters of error. The varied sample rate test shows that the algorithm can obtain position solutions down to a sample rate as slow as 50 MHz, with faster sampling rates having improved performance. The varied receiver geometry tests show that the algorithm can successfully position when receiver geometry becomes degraded, when the number of receivers is decreased leading to system observability degradation, and when an excess of information is provided to the system. The MUSCLE algorithm is then tested in a hardware-in-the-loop (HWIL) setup where simulated data is transmitted through one Universal Software Radio Peripheral (USRP) device and received by a second USRP. This HWIL data is then used for positioning with the MUSCLE algorithm and compared to the same environment fully in simulation. The HWIL test shows that real received data can arrive at a solution of less than 1 meter error for all signal sources. The simulation and HWIL tests validate that the MUSCLE algorithm successfully determined the position of multiple RF signal sources in short baseline environments.