I think it should be Cloud-Based VANET. As far as I can see, Involving Cloud in a real-time VANET applications (where response time should be very minimal) is not suggestible at this point. But for sure Cloud-Based VANET can be used to analyze the data (averaga speed of the vechicle, Usage of the sensors in the vehicle, response time of the sensors in the vehicles and Efficiency of the sensors. Communication time between the two vechiles and so on ... ) once it is collected by the Cloud. It can also used by the insurance companies for accident management. Just saying !!!
Raghu Avula, Thank you so much for your reply. I already wrote a paper about your last argument, "Vehicle Witnesses as a Service". You are absolutely right about Cloud-based VANET, but that is just one side of the picture where vehicles in VANET are using cloud infrastructure for different kinds of services, some of which you mentioned already. However, there is a buzz in research community about Datacenters and disposable supercomputers in parking lots where the resources of the vehicles are pooled together to do the job, perhaps on rent.
Well that depends what you actually want to achieve with VANET based cloud. In mobile cloud computing, the primary objectives are to achieve energy efficiency, enhanced computation power and execution support. In VANETs there is no energy issue, and if the nodes are equipped with powerful processors then the computational power enhancement and execution support also becomes useless. You need to think what you actually want to achieve with cloud powered VANETs.
@Atta ur Rehman Khan: Yeah I do agree with you in case of mobile cloud computing, but I would certainly have reservations about the computational power enhancement in vehicles. If you would consider the amount of data that is created by the vehicles on road, that would certainly be referred to as 'Big Traffic Data' according to the technical definition of big data. And for instance, dynamic traffic scheduling would require much more real-time sophisticated algorithms that a single processor in a vehicle either wouldn't be able to execute it, or take too long to execute it. There are many other applications of such concern, that would need the cars to pool their resources. However this scenario is in dynamic and short-term cloud federation. Another good example is the data center in the parking lot as aforementioned in my previous comment.