Vehicular-Computational Resource Geofencing: Efficient Spatiotemporal Uncertainty Estimation

Future vehicles are envisioned as smart objects equipped with sensors, actuators, storage capacity and computing resources. Intelligent transportation systems (ITS) have the potential to enhance transportation efficiency, road safety, and sustainability. Vehicular cloud computing (VCC) has emerged to facilitate intervehicle communication and data exchange enabling ITS services. However, existing VCC solutions are unable to satisfy the stringent quality of experience (QoE) requirements for running vehicular applications, especially delay- and throughput-sensitive applications, given the high communication delay and latency with the central cloud. To encounter these challenges, vehicular edge computing (VEC) solutions have emerged which rely on infrastructure or roadside units (RSUs) to bring cloud computing services closer to the user. Our objective is to enable the utilization of computational resources embedded in moving service vehicles to further enhance the QoE of vehicular users requesting computational tasks in urban environments. This methodology has the potential to extend the computation coverage to areas with limited RSU infrastructure as well as reduce the computational latency as services are provided closer to the user. To enable this approach, the service vehicles need to develop a spatiotemporal awareness of other computational resources within range and predict their future availability. The objective of this research is to propose an efficient computational resource geofencing solution to maximize the QoE of vehicular users requiring computational tasks. We evaluate the efficiency of the proposed resource geofencing algorithm under different realistic operating conditions. According to our evaluation, the service vehicles estimate the spatiotemporal uncertainty and predict the availability of computational resources with up to 99% confidence.