In this work, an intelligent mobile edge computing (MEC) network is studied for Internet of Things (IoT) in the presence of eavesdropping environments, where there are multiple users who can offload their confidential tasks to the computational access point (CAP) for the assistance of computation. One unmanned aerial vehicle (UAV) attacker exists in the system and it can listen to the confidential data transmission from the users to the CAP. We optimize the system design of the intelligent MEC network, by adaptively allocating the offloading ratio and wireless bandwidth, to reduce the linearly weighted cost of the latency as well as energy consumption (EnC). Specifically, starting from the deep reinforcement learning, we devise a deep Q-network (DQN) network to adjust the offloading ratio and transmission bandwidth, which can help calculate the computational tasks and suppress the eavesdropping from the UAV efficiently. We finally provide some simulation results to validate the proposed offloading strategy. In particular, the proposed offloading strategy can achieve a much lower cost compared to the conventional ones, in the terms of latency and EnC.INDEX TERMS Deep reinforcement learning, Internet of Things, mobile edge computing, task offloading, unmanned aerial vehicles.
In this paper, we investigate intelligent mobile edge computing networks (MEC) with multiple computational access points (CAPs) for Internet of Things (IoT), where the user with limited computational capability has some tasks to be computed with the help of the CAPs. We use the communication and computational latency to measure the system outage probability. In order to enhance the system performance, two CAP relay selection criteria are proposed, where one is based on the computational capability at the CAPs while the other criterion is based on the instantaneous channel parameters from the user to the CAPs. To further improve the system performance, the offloading ratio is optimized through minimizing the system outage probability, which determines how many parts of tasks to be computed at the CAPs. Numerical and simulation results are finally presented to verify the effectiveness of the proposed CAP selection criteria and offloading ratio strategy.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.