Accuracy Improvement of RFID based 2D Tracking and Localisation

The purpose of localization and tracking technology in indoor application is to extract moving object parameters accurately and precisely. This thesis investigates the problem of how to utilize RFID technique for the accurate and precise extraction of indoor 2D moving object position parameters. Firstly, a framework named RFID-Loc with three modules: RFID-Loc Infrastructure, RFID-Loc Data Filter and RFID-Loc Localisation Algorithm, is established from a theoretical perspective. This framework can guide the research and design of methods used in an RFID based object localisation system with enhanced localisation accuracy and precision. Secondly, from practical perspective, few methods are proposed in RFID-Loc framework to improve the localisation accuracy and precision. A sparse RFID Tag Arrangement strategy is proposed in this RFID-Loc framework, aiming at reducing the impacts of regular false reading error from RFID infrastructure level on localisation precision. The efficiency of this methods and the assumptions upon which it relies, are investigated empirically. A rectangle-based feature selection method is justified as the major RFID Data Filter algorithm, with the capability of maximally reducing regular false reading errors. The possibility to resist unexpected false reading error in an RFID-Loc system is investigated by discussing and comparing several RFID-based localisation algorithms. A dynamic localisation algorithm for RFID-Loc system is proposed to accurately and precisely extract moving object position parameters overtime in an RFID-Loc system. This algorithm is shown to have a better capability of resisting unexpected false reading error than conventional localisation algorithms used in RFID-based localisation systems, while having a higher computational complexity. By following the theoretical guidelines in RFID-Loc framework and implementing the proposed methods, the experimental results demonstrate that the localisation accuracy and precision can be significantly improved, up to 10 centimetres and 3 centimetres under current RFID devices