Vehicle positioning utilising radio frequency identification devices with geo-located roadside furniture upon urban-roads

Vehicle positioning technology currently is one of the popular research topics in telecommunication engineering. With advances in the Vehicular Ad-hoc Network (VANET), Driverless Cars (DCs), the Intelligent Transport System (ITS), the Internet of Things (IoT) have placed greater demands on traditional positioning technology. Typically, to make better driving decisions, vehicle control systems of DCs require centimetre-level positioning accuracy.

However, only using the Global Navigation Satellite System (GNSS) is not accurate enough to match the required positioning levels because of signal attenuation and the

underlying latency between the satellite and receivers. Therefore, multi-model positioning systems, combining GNSS, light detection and ranging (LiDAR), Inertial Measurement Units (IMUs), Artificial Intelligence (AI), and Cloud Computing, are utilised for autonomous vehicles at centimetre-level positioning accuracy.

In order to overcome the above-mentioned shortcomings of conventional positioning systems, a new and timely vehicle positioning framework, named radio-frequency identification - driverless car positioning system (RFID-DCPS), is proposed in this thesis as a complement to existing multi-model positioning systems. This system is designed based on a local database of accurately geo-located Radio Frequency Identification (RFID) transponders on roadside furniture near roads and car-based interrogators. When the

vehicle drives past the tagged roadside furniture, the interrogator mounted on the vehicle interrogates the location of the tagged roadside furniture within the reading range and determines the real-time positioning of the vehicle.

In this thesis, the research topics can be divided into five parts: 1) a survey and an overview of RFID-DCPS, as well as introductions of the current positioning methods and

autonomous vehicles; 2) roadside furniture locations collection and its deviation estimation; 3) an intelligent driving model; 4) multiple access schemes dealing with anticollision (transponder-to-transponder and interrogator-to-interrogator) in RFID; and 5) an RFID channel evaluation and positioning algorithm (3D-TDOA) based on roadside furniture collision and deviation estimation caused by the velocity. The first part consists of the introduction of RFID-DCPS and the recent development of autonomous vehicles and corresponding multi-model positioning methods. As for the second contribution, a selected route is surveyed in Loughborough’s town centre using a modified intelligent model and a SPYGLASS app regarding location data of roadside furniture and its deviation estimation. The third one indicates the applied contribution of an intelligent model that can simulate vehicles’ position and movement on the measured circular track around Loughborough town. The novel work of 3) and 4) concentrates on the issue of anti-collision, both transponder-to-transponder and interrogator-to-interrogator, and proposes new methods to deal with multiple access schemes. Combined with the position data of roadside furniture, multiple access schemes and ignoring the latency of 4G or 5G uplink and downlink, the positioning estimation of RFID-DCPS is regarded as the fifth contribution.

The results from the simulation demonstrate that it is feasible to consider the proposed positioning system as the component of existing multi-model positioning systems. It

should also be noted that the deviations in RFID-DCPS and the Doppler errors are mainly caused by the triangulation algorithm and velocity, respectively. The average positioning

accuracy in Loughborough urban roads tends to be 7.49 m under the vehicle density of 50, and the highest velocity is 30 m/s in the interrogation range.