Ticket-Based Traffic Flow Control at Intersections for Internet of Vehicles

Abstract

This paper proposes a traffic flow control system at intersections for Internet of Vehicles by combining ideas from process scheduling in Operating Systems with a ticket mechanism. The paper models an intersection as a network of mutual exclusive resources and the vehicle flows as processes that access the resources. Two graph models: Flow Intersection Network and Flow Resource Allocation Network are introduced as intermediate representations to derive which traffic flows can run concurrently and how long they take. Based on this knowledge, the ticket based traffic control system uses a message protocol and two asynchronous processes: flow scheduler and flow coordinator to admit, schedule and coordinate concurrent vehicles through the intersection without collisions or deadlocks. Three classic scheduling algorithms: First Come First Serve (FCFS), Shortest Service Time Next (SSTN), and Highest Response Ratio Next (HRRN), are incorporated into the scheduler. The experiments on large scale simulated traffic data show the scheduling algorithm scales linearly with the number of traffic flows and vehicles. The tests also show that at low to medium traffic loads, the throughput and delay of the three algorithms are mixed, but at heavy traffic loads, SSTN and HRRN outperform the commonly used FCFS.