Soca: secure offloading considering computational acceleration for multi-access edge computing

Abstract

Due to the complexity and security requirements of edge computing environments and the limited resources of terminals, secure offloading in multi-access edge computing (MEC) networks has emerged as a critical and urgent research area. However, many studies on task offloading often ignore the necessary balance between security requirements and efficiency. To address this issue, we propose a Secure Offloading Strategy Considering Computational Acceleration, named SOCA, designed to bolster security while preserving offloading efficiency. Specifically, the secure offloading problem is modeled as a multi-objective optimization problem by achieving a composite function of latency mitigation and security metrics as the optimization objective, which is solved by the ChaCha20-based offloading decision algorithm (ChaCha20-ODA). The algorithm employs the ChaCha20 encryption protocol as its security mechanism. By executing a quarter-round function to generate a keystream, it provides robust protection for data tasks, ensuring that the data remains impervious to malevolent interception by adversaries throughout the transmission process. Furthermore, to improve the computational efficiency of task offloading, the algorithm simultaneously leverages both edge and local computing resources, achieving computational acceleration by optimizing the appropriate offload ratio. The experimental results illustrate that as compared with baselines, our approach achieves remarkable improvement in the balance between latency and safety benchmarks, which demonstrates the superiority of our method.