Dynamic service migration and resource management for vehicular clouds

Vehicular cloud computing has received huge attention in business and scientific communities, which integrates two emerging fields, namely cloud computing and vehicular ad hoc networks. It acts as a data center by using the underutilized resources of the networked vehicles. Moreover, many studies suggest these vehicles as potential candidates for hosting virtual machines (VMs). As a result, a VM can be set up using the vehicular resources, and it can be transferred from one vehicle to another vehicle to continue its execution, under some circumstances. It enables the vehicular environment to provide services to the user requests, that are submitted to the cloud. However, the mapping of such requests to the VMs (or hosted vehicles) and service migration is very much challenging, and not well-studied in the literature. In this paper, we propose a dynamic service migration algorithm for vehicular clouds. The algorithm consists of three phases, estimation, assignment and migration. The performance is carried out through simulation runs using two scenarios of six datasets, and compared with three well-known algorithms, namely vehicular VM migration-uniform, round robin and mobility and destination workload aware migration using four performance measures. The comparison results followed by statistical validation using T test show the superiority of the proposed algorithm over the existing algorithms.