Date of Award
Master of Science (MS)
Vehicular networks are commonplace, and many applications have been developed to utilize their sensor and computing resources. This is a great utilization of these resources as long as they are mobile. The question to ask is whether these resources could be put to use when the vehicle is not mobile. If the vehicle is parked, the resources are simply dormant and waiting for use. If the vehicle has a connection to a larger computing infrastructure, then it can put its resources towards that infrastructure. With enough vehicles interconnected, there exists a computing environment that could handle many cloud-based application services. If these vehicles were electric, then they could in return receive electrical charging services.
This Thesis will develop a simple vehicle datacenter solution based upon Smart Vehicles in a parking lot. While previous work has developed similar models based upon the idea of migration of jobs due to residency of the vehicles, this model will assume that residency times cannot be predicted and therefore no migration is utilized. In order to offset the migration of jobs, a divide-and-conquer approach is created. This uses a MapReduce process to divide the job into numerous sub-jobs and process the subtask in parallel. Finally, a checkpoint will be used between the Map and Reduce phase to avoid loss of intermediate data. This will serve as a means to test the practicality of the model and create a baseline for comparison with future research.
"Supporting Big Data at the Vehicular Edge"
(2018). Master of Science (MS), thesis, Computer Science, Old Dominion University, DOI: 10.25777/6pw2-7545