Key points are not available for this paper at this time.
Mobile edge computing (MEC) pushes computing and caching resources close to the network edge, which allows devices to offload computation-intensive tasks to MEC servers. Considering that wireless signals may be easily blocked by obstacles, reconfigurable intelligent surface (RIS) has emerged as a promising technique to improve the efficiency of computation offloading. In this paper, we consider a RIS-assisted MEC network, where a MEC server caches service programs required for task execution and a RIS helps computation offloading in the presence of eavesdropping. Due to the diversity of services and the broadcast nature of wireless channels, it is challenging to achieve efficient and secure computation offloading in this network. Therefore, we first formulate a task completion delay minimization problem by jointly optimizing service caching, computation offloading decisions, RIS passive beamforming, and transmit power subject to the constraints of secure offloading rate and limited storage space. To address the highly non-convex nature of the problem, we then develop a dual-layer optimization algorithm via a vertical decomposition on its layered structure. The outer-layer problem, which deals with service caching and computation offloading decisions, is solved by a cross-entropy-based caching and offloading learning algorithm. For the inner-layer problem that optimizes RIS passive beamforming and transmit power, we utilize a horizontal decomposition by invoking the block coordinate descent method. Finally, simulation results demonstrate that the proposed scheme exhibits performance improvements compared to several baseline schemes.
Wu et al. (Fri,) studied this question.