Mesh Network Smart UWB Indoor Positioning Tracking Light P2P System

Ko, Zi Yuan (2024) Mesh Network Smart UWB Indoor Positioning Tracking Light P2P System. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text KO ZI YUAN- Full Text.pdf Restricted to Registered users only Download (14MB)

Abstract

This final year project aimed to enhance the performance of the Smart UWB Indoor Positioning Tracking Light System by integrating a mesh network with a peer-to-peer (P2P) communication system. The project focused on three main objectives: mitigating delays caused by the centralized architecture of the smart indoor positioning lighting system, overcoming communication range limitations in larger indoor spaces, and building and validating the hardware of the mesh network smart UWB indoor positioning tracking light P2P system. Additionally, the project sought to improve the precision of the Indoor Positioning System (IPS) through machine learning techniques. The results focused on system response time performance, system range and coverage improvement, and Indoor Positioning System (IPS) accuracy. System response time performance was significantly improved with the implementation of the P2P system, reducing latency by 79.94% compared to the previous centralized architecture. This improvement was attributed to streamlining backend processes and leveraging Bluetooth Low Energy (BLE) technology. Furthermore, the adoption of a BLE mesh network architecture extended the system's operational range and facilitated reliable communication between devices within the network. The project also addressed IPS accuracy through machine learning techniques, particularly for Line of Sight (LOS) and Non-Line of Sight (NLOS) classification. Various machine learning algorithms were evaluated, and results indicated promising accuracy levels, with room for further improvement. Additionally, distance calibration regression models were trained and tested, with Ridge Regression demonstrating superior performance compared to other machine learning models and polynomial regression calibration (PRC) models. Overall, the project achieved the objectives of enhancing system performance, extending operational range, and improving IPS accuracy through the integration of mesh networking, P2P communication, and machine learning techniques. These advancements lay the foundation for future developments in smart indoor positioning systems, offering enhanced precision and reliability in indoor tracking applications.