Fuzzy traffic signal control in vehicular ad hoc networks

Traffic signal control (TSC) is a major concern for big cities in the world. Inefficient and ineffective intersection control arise economic, environmental and social problems. It relies mostly on sensors, such as loop detectors and cameras which are costly and limited to sensor capability. Vehicular ad hoc networks (VANETs) enable the TSC to acquire real information for vehicles. This requires road side unit (RSU) in the centre of an intersection capable of communicating with vehicles. TSC's with sensors have two major drawbacks, installation/maintenance costs and rate of failure. On the other hand, recently TSC's with VANETs have also two major drawbacks, sophisticated traffic surveillance and ineffective control solution. The main reason for that is the low number of vehicles that use this technology, so called penetration rate (PR). This thesis focuses on recently TSC algorithms for signalized intersections under VANET environment under low PR. This is done through VANET communications, simple traffic surveillance and intelligent control solution. In VANET communications, vehicle-to-everything (V2X) communications are particularly well-suited for traffic surveillance. This is due to their low latency and their ability to communicate instantly between vehicles via vehicle-to-vehicle (V2V) communication and between vehicles and infrastructure via vehicle-to-infrastructure (V2I) communications. Simple traffic surveillance requires base information for vehicles (such as position and speed). Because of vehicles dynamic behaviour, these information changes continuously and becomes fuzzy under low PR. In this context, this thesis proposed an accumulative information approach based on fuzzy logic. Two approaches investigated in this field, vehicles approaching and leaving pattern and traffic delay model. The results encourage us to use traffic delay model for intelligent control. Recent TSC solutions focus on self-organizing algorithms (e.g. platoon, phase, marching and congestion based). The main idea of each algorithm considered traffic as an adaptive rather than an optimizing problem. Based on the local information, this problem can be solved. The major drawback for each one, it is suitable for specific traffic condition and not for others. In order to solve this issue, two approaches have been proposed to select one suitable algorithm based on traffic condition criteria: simple logic and fuzzy logic. The two proposed approaches have been simulated and evaluated with behaviours of recently developed algorithms in this field. The results indicate good performance of our TSCs (more specifically fuzzy logic) using VANET environment even under low PR.