DPS-CR: A Task Scheduling Algorithm Based on Computation Reuse in Vehicular Edge Computing

In the realm of vehicular edge computing (VEC), where tasks generated by vehicles can choose local processing or be offloaded to nearby edge servers. However a prevalent issue arises due to the inherent spatiotemporal correlation observed among these tasks, this correlation results in similar tasks producing identical output results, leading to a significant amount of redundant computation. Consequently, this redundancy not only increases the workload on edge servers but also prolongs task completion time. Therefore, we design the Dynamic Programming Scheduling algorithm based on Computation-Reuse (DPS-CR), which adopts a computational reuse model that relies on vehicle road data fusion, while solving the offloading problem of vehicle generation tasks with bandwidth resource requirements under the common coverage range of multiple base stations (BSs). We simulated the situation of vehicles passing through the intersection in the uncongested scenario and the congested scenario. The experimental results show that, compared with baselines, the DPS-CR can significantly reduce system cost (SC) by effectively utilizing the resources offered by BSs. Moreover, our proposed DPS-CR can minimize the probability of task loss in vehicular environments.