Joint Optimization of Task Offloading and Resource Allocation Based on Differential Privacy in Vehicular Edge Computing

Abstract

In the Internet of Vehicles (IoVs), task offloading is necessary to ensure the low-response delay due to the limitation of vehicular computational capacity. Task offloading involving social behavior can improve the utilization of computational resources in IoVs. To offload tasks effectively, the connected vehicles (CVs) need to upload context information, such as speed and location to road side unit (RSU) and base station (BS), which brings dramatic threat and risk for CVs’ privacy security. To solve the above-mentioned issue, we propose a privacy-preserving vehicular edge computing (PP-VEC) system architecture in this article. In the PP-VEC, the vehicular tasks can be offloaded to RSUs and adjacent CVs with adequate computing resources. Privacy mechanism disturbs the context information of CVs based on differential privacy technology before uploading it to the BS for offloading decisions to protect the CVs’ privacy. This article adopts the local differential privacy algorithm based on histogram algorithm and proposes a K-neighbor joint optimization of task offloading and resource allocation algorithm (K-NJTA) to optimize the global delay of task execution. We demonstrate the effectiveness of the proposed methods by simulation experiments. The results demonstrate K-NJTA on task execution delay and our local differential privacy algorithm can protect CVs’ privacy while has less effect on the task offloading algorithm due to the context information distribution.