Dynamic Task Scheduling with Mobility Consciousness in Vehicular Fog Computing Environments

While Vehicular Fog Computing (VFC) is a hot research area for boosting processing power in connected vehicles (Internet of Vehicles), efficiently assigning the best resources is tricky. This is because new services like augmented reality and self-driving cars are exploding in popularity. These services require lightning-fast responses, powerful computing, and all while dealing with constantly moving vehicles. The biggest difficulty is guaranteeing tasks are completed super-fast (strict latency) as vehicles move in and out of range of roadside stations (RSUs). This research tackles the challenge of scheduling tasks for connected vehicles. It considers both how long a task takes to complete (latency) and how much battery power it uses (energy consumption). The system makes decisions about whether to run the task on the vehicle itself, a nearby roadside station (RSU), or even another RSU further down the road based on the vehicle's planned route. This approach aims to find a balance between getting tasks done quickly and keeping the vehicles running for as long as possible. Initially, we use a special technique called Markov renewal process (MRP) to capture how vehicle moves around over time. This method helps us understand the vehicle's mobility patterns. We then prove some technical properties about our goal of minimizing both latency and energy consumption. This lets us develop a fast and effective method (greedy heuristic) to find a good solution. On top of that, we propose a more powerful approach that combines two existing optimization techniques (Moth-flame and particle swarm) to tackle scheduling tasks in very large networks with moving vehicles. We tested our new scheduling method against existing ones, and computer simulations showed it to be very effective in solving task scheduling problems. The simulation outcomes demonstrate that our suggested MFO-PSO approach reduces average delays in tasks by 23.4% and uses 17.8% less energy than typical scheduling methods.