MAVNet: A Mobile Agent based framework for Vehicular Networks

The advent of VLSI technologies has provided a breakthrough in IC manufacturing processes leading to a substantial increase in the population of a low footprint and low power mobile computing devices. A network of such devices that can communicate with one another can aid in the realisation of a ubiquitous and pervasive computing ecosystem. If these participating devices are vehicles, then they form a particular class of mobile peer-to-peer network categorised as Vehicular Ad-hoc Networks (VANET). The devices which mainly include cars, are highly mobile thus posing numerous challenges in interdevice communication, synchronisation and sharing of data. Monitoring vehicular and battery health, movement, identification, etc. on the run are challenges that need to be addressed. In this paper, we describe an attempt to transform a vehicle into a mobile computing node that can host a multi-agent platform. With static agent(s) resident on these platforms and mobile agents flitting across these vehicles, we explain how a Road Side Unit (RSU) can track and obtain information about the traffic. Since these agents are autonomous pieces of static or mobile software, it is not only possible for them to obtain information about the vehicle but also gain control over it, if required. Further, since mobile agents can migrate from one vehicle to another, they possess the ability to aggregate or disseminate information on the status of the traffic. In this paper, we discuss MAVNet, a mobile agent-based framework for vehicular networks where we propose the deployment of mobile agents for data acquisition, on-board processing and sharing within a network of vehicles and the infrastructure based RSUs. As a proof-of-concept, a partial real-world deployment along with an experimental have been discussed in the paper.