Abstract
In recent years, vehicular cloud computing (VCC) has gained vast attention for providing a variety of services by creating virtual machines (VMs). These VMs use the resources that are present in modern smart vehicles. Many studies reported that some of these VMs hosted on the vehicles are overloaded, whereas others are underloaded. As a circumstance, the energy consumption of overloaded vehicles is drastically increased. On the other hand, underloaded vehicles are also drawing considerable energy in the underutilized situation. Therefore, minimizing the energy consumption of the VMs that are hosted by both overloaded and underloaded is a challenging issue in the VCC environment. The proper and efficient utilization of the vehicle's resources can reduce energy consumption significantly. One of the solutions is to improve the resource utilization of underloaded vehicles by migrating the over-utilized VMs of overloaded vehicles. On the other hand, a large number of VM migrations can lead to wastage of energy and time, which ultimately degrades the performance of the VMs. This paper addresses the issues mentioned above by introducing a resource management algorithm, called resource utilization-aware VM migration (RU-VMM) algorithm, to distribute the loads among the overloaded and underloaded vehicles, such that energy consumption is minimized. RU-VMM monitors the trend of resource utilization to select the source and destination vehicles within a predetermined threshold for the process of VM migration. It ensures that any vehicles' resource utilization should not exceed the threshold before or after the migration. RU-VMM also tries to avoid unnecessary VM migrations between the vehicles. RU-VMM is extensively simulated and tested using nine datasets. The results are carried out using three performance metrics, namely number of final source vehicles (nfsv), percentage of successful VM migrations (psvmm) and percentage of dropped VM migrations (pdvmm), and compared with threshold-based algorithm (i.e., threshold) and cumulative sum (CUSUM) algorithm. The comparisons show that the RU-VMM algorithm performs better than the existing algorithms. RU-VMM algorithm improves 16.91% than the CUSUM algorithm and 71.59% than the threshold algorithm in terms of nfsv, and 20.62% and 275.34% than the CUSUM and threshold algorithms in terms of psvmm.