Employing Ray-Tracing and Least-Squares Support Vector Machines for Localisation

Area of Science:

• Signal processing
• Machine learning
• Wireless communication

Background:

• Multipath environments pose significant challenges for accurate device localisation.
• Existing localisation schemes often require complex non-line-of-sight (NLOS) identification and mitigation.
• Support vector machines offer a robust framework for classification and regression tasks.

Purpose of the Study:

• To evaluate the efficacy of least-squares support vector machines (LS-SVM) for 2-D localisation in multipath environments.
• To compare the performance of an LS-SVM-based direct localisation method against traditional time difference of arrival (TDOA) and time of arrival/angle of arrival (TOA/AOA) schemes.
• To assess the applicability of the proposed method in both line-of-sight (LOS) and NLOS conditions without explicit NLOS handling.

Main Methods:

• Utilisation of ray-traced data for training and testing the localisation algorithms.
• Implementation of a direct localisation scheme employing LS-SVM.
• Comparative analysis against TDOA and TOA/AOA methods, including scenarios with and without NLOS.
• Evaluation based on outage probability under varying environmental conditions.

Main Results:

• The LS-SVM direct method demonstrates superior outage performance compared to TDOA and TOA/AOA in NLOS environments.
• In LOS environments, TDOA exhibits better outage performance.
• For outage probabilities of 20% or greater, TOA/AOA performs competitively, but the LS-SVM direct method becomes increasingly advantageous as outage probability decreases.

Conclusions:

• The LS-SVM direct localisation method provides a robust and effective solution for multipath environments, particularly in NLOS conditions.
• The proposed scheme eliminates the need for explicit NLOS identification and mitigation, simplifying the localisation process.
• The method's performance is competitive and often superior to existing techniques, especially at lower outage probabilities, making it a promising approach for advanced localisation systems.