Exploring Computational Energy Loss in Connected Vehicles
With advancements in connected vehicle technology, sophisticated computing equipment is installed to assist resource-intensive applications for better and faster processing. However, due to the high demand for computation, local resources are insufficient, and therefore, tasks are offloaded to nearby network edges to meet task deadlines. A similar approach is adopted for vehicle-to-vehicle task offloading, where underutilized vehicles are used to meet the computation demands of heavily loaded vehicles. Due to dynamic changes in topology caused by vehicle speed and direction, many tasks fail to deliver results after remote computation. In this work, we explore energy consumption in the latter approach, where tasks are executed but fail to deliver results. Furthermore, we propose a multi-layer, energy-enabled task offloading strategy that relies on degree, closeness, and betweenness centrality as the initial selection mechanism, where the second tier selection relies on features such as service time and potential path diversion time. The results show a 56% to 45.17% energy loss reduction in the proposed approach with varying vehicle arrival rates.