Communication-wise Comparison of the Online Resource Allocation Methods in CAV Fleets

Abstract

Deploying fleets of Connected Autonomous Vehicles (CAVs), with limited peer-to-peer communication ranges, in order to provide On-demand Transport (ODT) services, requires a careful choice and evaluation of solution methods for resource allocation problems. We adopt in this paper a multiagent approach consists of: 1) defining a generic model to ODT's dynamic resource allocation problem in connected autonomous vehicle fleets, taking into account the limited connectivity and communication constraints, 2) behavior abstraction of AV Agents, and 3) abstracting the solution methods as Coordination Mechanisms (CMs) to define the characteristics of a solution method and its requirements to implement the corresponding planning sub-behavior. Using this methodology, we were able to compare the performance of a variety of solution methods with a set of evaluation criteria, namely for the solution Quality of Service (QoS) and Quality of Business (QoB). In this work, we focus on the communication-wise evaluation criteria, such as connectivity and network load.