Optimizing Resource Allocation for Vehicular Edge Computing: A Hybrid Approach for Enhanced Efficiency and Task Management

Abstract

In this research, present a hybrid optimized resource allocation model that takes into account multiple factors, including the status of available resources, geographical distances, and available network bandwidth, and the specific requirements of tasks to be executed with in Vehicular Edge Computing (VEC). The operational flow of the model begins with network formation, creating the foundation for effective communication and resource sharing. Feature extraction techniques are then applied to obtain relevant information from the available resources in the VEC domain. Following feature extraction, a detailed analysis identifies all connectable vehicles within the VEC. Feature selection processes are underpinned by the combination of the Walrus Optimization Algorithm and the Osprey Optimization Algorithm (WaOA-OOA) which refine the design, enabling a more efficient allocation. The core of the model involves the computation of a task-vehicle cost-time matrix, incorporating factors such as resource capabilities and task requirements. This matrix guides the subsequent vehicle clustering process, which utilizes the Farthest First K-means algorithm to form clusters of vehicles with similar resource profiles. The task allocation phase optimally assigns tasks to vehicle clusters, ensuring efficient resource utilization and minimal task delay. The Drawer Algorithm (DA) is proposed to optimize resource allocation by minimizing task completion times. In this algorithm, vehicles are grouped for each task, and the goal is to determine the most efficient way to assign subtasks to vehicles while minimizing the overall task completion time. The algorithm uses a cost-delay matrix to represent the time it takes for each subtask to be completed on each candidate vehicle where the key idea is to divide the subtasks among the vehicles in a way that optimizes the system's overall performance, which is formulated as a 0-1 integer linear programming problem. The proposed model seamlessly integrates with cloud servers to securely save allocation results, making them accessible for subsequent task execution. Simulation results underscore the effectiveness of our proposed method, revealing significant reductions in vehicle power consumption while consistently meeting task delay requirements. This research advances the resource utilization in both cloud and edge computing domains, and advances the state of resource allocation in VEC.