Vehicular clouds is an active area of research that has emerged at the nexus of conventional cloud computing and vehicular networks. The defining differences between conventional and vehicular clouds include the heterogeneity and volatility of compute resources and the bandwidth-challenged network fabric. A variety of new architectures and services for vehicular clouds have been proposed, mostly as incremental extensions of the VANET platform. As vehicular cloud research continues and expands, a careful eye should be kept on the restrictions that come with the mobility, limited network, and heterogeneity of resources. The first main contribution of this chapter is to survey recent work of VCs with an eye on the realistic and unrealistic. Our second main goal is to realign the VC community with a realistic vision for the future by spelling out a number of challenges faced by the VC research community.
Top1. Introduction
The vision and promise of vehicular networks, populated by vehicles that can communicate to improve safety and provide traveler information, has captivated the networking research community over the past two decades. Applications and protocols have been proposed, standards were and are still being written, and hardware is being developed to support various applications. One of the turning points that we have witnessed in the past decade was the emergence of self-driving cars. Several manufacturers have already committed resources towards large-scale production of such self-driving cars. This trend will drive the production of vehicles equipped with advanced computational, communication, storage, and sensing capabilities. With close to seven million new cars purchased in the US alone each year, we will soon have many millions of these advanced vehicles on our roadways and city streets.
Many of these advanced vehicles will spend hours at a time in a parking garage, parking lot, or driveway. While parked, or while crawling in a traffic slowdown, the computing and storage resources of these vehicles will be untapped and the opportunity for their use will be wasted. Given the right incentives, the owner of a vehicle may decide to rent out excess on-board capabilities. Owners of parking garages may decide to provide such incentives for vehicle owners to allow the use of their parked vehicles. For example, future airports will power travelers' vehicles to harness their computing resources, allowing for on-demand access to parking lot data centers.
Likewise, vehicles crawling in a traffic slowdown will be able to donate their compute resources so that municipal traffic management centers can run complex simulations designed to help alleviate the effects of congestion by city-wide rescheduling of traffic lights. We also anticipate that in emergency evacuation situations, the combined compute power of the vehicles will allow evacuation management centers to run sophisticated simulations in close to real-time in order to effectively manage the evacuation event.
To enable such a future, Eltoweissy et al. (Eltoweissy, Olariu, & Younis, 2010) introduced and developed the concept of a Vehicular Cloud (VC) defined to be:
A dynamic group of vehicles whose excess computing, communication, storage, and sensing resources are coordinated and dynamically allocated to authorized users.
As it turns out, VCs are a non-trivial extension of conventional cloud computing. The distinguishing characteristic of VCs is mobility, which results in a dynamically changing amount of available resources. It is therefore clear that the VCs can only achieve their full potential if their basic architecture is tailored to offer a seamless integration of the cyber-physical resources of the participating vehicles. In particular, the architecture must adapt its managed vehicular resources allocated to an application according to dynamically changing requirements and systems conditions.
While some flavors of VCs may be run much like conventional clouds, their novelty is that they can also be used to supplement a municipal separtment of transportation's computing resources during extended periods of traffic congestion or even during emergency conditions, such as an evacuation (Abuelela & Olariu, 2010).
In such scenarios, the VC can provide on-demand solutions to traffic events that have occurred but whose timely resolution cannot be met reasonably with pre-assigned assets or in a proactive fashion. It is essential to investigate issues related to dynamic resource management in a VC -- a challenging, dynamically varying wireless environment. The results obtained for VCs could be applied in the development of other transformative mobile computing applications.
The first main goal of this chapter is to survey recent work of VCs with an eye on the realistic and unrealistic. Our second main goal is to realign the VC community with a realistic vision for the future by spelling out a number of challenges faced by the VC research community.