The Global Positioning System (GPS) is widely used for outdoor positioning and navigation. However, GPS relies on satellite signals that are significantly attenuated by buildings and obstacles, making it unreliable indoors. As a result, alternative approaches are required to enable accurate device positioning in scenarios with multipath interference and signal attenuation.

This project aims to address the limitations of GPS by developing a machine learning (ML)-based positioning prediction model that operates independently of satellite signals and instead leverages communication between devices within the same 5 GHz localised network. To achieve this, the project will evaluate and compare signal-based, direction-based and time-based indoor positioning methods under emulated conditions. RSSI, TDOA, and DOA/AOA measurements are generated under indoor and outdoor scenarios with LOS/NLOS link states, obstacle interactions, and measurement noise. Multipath propagation is not explicitly simulated in this implementation; instead, obstruction effects are approximated through LOS/NLOS classification, blocker counts, environment-dependent noise, and optional RSSI obstacle attenuation. The specific problem addressed in this project is whether a machine learning model can outperform or refine conventional indoor positioning methods by learning patterns caused by environmental conditions in both indoor and outdoor propagation environments.

This problem provides a sufficient challenge for an undergraduate dissertation due to the mathematical complexity of using device positioning methods and the intricacies of generating realistic network behaviours. Furthermore, the project requires evaluating multiple positioning methods, as well as designing and training a machine learning model. Hence, this project will demonstrate appropriate technical depth, independent problem-solving, and critical analysis for a computer science dissertation.

This project is sponsored by Clear-Com, a telecommunications company that provides hardware and embedded software solutions for a range of professional sectors, including entertainment, nuclear facilities, military operations, and space exploration. Since Clear-Com devices are used in both indoor and outdoor environments, where accurate device positioning can be crucial, they can particularly benefit from this project. However, many organisations face the same issues and constraints, allowing for the outcomes of this project to be applicable beyond a single company or use case.