Joint Service Caching and Secure Computation Offloading for Reconfigurable-Intelligent-Surface-Assisted Edge Computing Networks

• Published in: IEEE internet of things journal Vol. 11; no. 19; pp. 30469 - 30482
• Main Authors: Wu, Mengru, Chen, Weijin, Qian, Liping, Guo, Lei, Lee, Inkyu
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.10.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Array signal processing; Beamforming; Caching; Computation offloading; Decisions; Decomposition; Delays; Edge computing; Entropy (Information theory); Internet of Things; Machine learning; Mobile computing; Mobile edge computing (MEC); Optimization; reconfigurable intelligent surface (RIS); Reconfigurable intelligent surfaces; secure computation offloading; Security; Servers; service caching (SC); Task analysis
• ISSN: 2327-4662; 2327-4662
• DOI: 10.1109/JIOT.2024.3404972

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 article, 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 (SC), 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 nonconvex 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 SC 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.