Optimal Traffic Scheduling in Vehicular Delay Tolerant Networks

In a vehicular delay tolerant network, the problem of scheduling the traffic from a roadside unit to passing-by vehicles is studied. A cost function is associated with a scheduled transmission. And if a delay bound is violated, a penalty cost is charged. The optimal stopping theory is applied to decide when the roadside unit should schedule its accumulated traffic to a passing-by vehicle such that the rate of cost is minimized. It is shown that the optimal scheduling strategy is a pure-threshold strategy, i.e., upon a vehicle arrival, if the queueing delay at the roadside unit is above a threshold, it is optimal to schedule a transmission via the vehicle. Simulation results validate the effectiveness of the derived optimal rule.

I. Introduction

Vehicular delay tolerant networks have recently gained much research attention [1]–[4]. A typical vehicular delay tolerant network consists of unconnected stationary RoadSide Units (RSUs) and is deployed along highways in sparsely populated or unpopulated zones. It is very likely that the RSUs in those zones are driven by batteries [3]. Only a few RSUs are connected to the Internet, and other RSUs should send their traffic to those RSUs with Internet connection. There is no direct connection among the RSUs, and passing-by vehicles serve as relays, to help forward traffic among the RSUs. For such networks, delay minimization problem in a Markov decision process framework is studied in [1], while a model is given in [2] to probabilistically decide whether a passing-by vehicle is suitable to carry the traffic of an RSU such that the total transit delay is minimized. Minimizing the energy consumption of the RSUs, under the constraint of satisfying the communication requirements of the passing-by vehicles, is studied in [3].