This letter considers a multi-access mobile edge computing (MEC) network consisting of multiple users, multiple base stations, and a malicious eavesdropper. Specifically, the users adopt the partial offloading strategy by partitioning the computation task into several parts. One is executed locally and the others are securely offloaded to multiple MEC servers integrated into the base stations by leveraging the physical layer security to combat the eavesdropping. We jointly optimize power allocation, task partition, subcarrier allocation, and computation resource to maximize the secrecy offloading rate of the users, subject to communication and computation resource constraints. Numerical results demonstrate that our proposed scheme can respectively improve the secrecy offloading rate 1.11%--1.39% and 15.05%--17.35% (versus the increase of tasks' latency requirements), and 1.30%--1.75% and 6.08%--9.22% (versus the increase of the maximum transmit power) compared with the two benchmarks. Moreover, it further emphasizes the necessity of conducting computation offloading over multiple MEC servers.
翻译:本信考虑了由多个用户、多个基站和一个恶意窃听器组成的多存取移动边缘计算(MEC)网络。 具体来说, 用户采用部分卸载策略, 将计算任务分成几个部分。 一个是本地执行的, 其他则安全地卸载到多个MIC服务器中, 利用物理层安全来打击偷听。 我们共同优化了电力分配、 任务分割、 子载体分配和计算资源, 以最大限度地降低用户的保密卸载率, 但须受通信和计算资源限制。 数字结果显示, 我们提议的计划可以分别改进1. 11%- 1. 39 % 和 15. 05- 17.35% ( 任务耐久要求的增加) 的保密卸载率, 以及1. 30- 1. 75% 和 6. 08%- 9. 22% ( 最大传输功率的增加) 与两个基准相比。 此外, 它进一步强调必须在多个MEC 服务器上进行计算卸载的卸载率。