Internet of Things (IoT) expects to incorporate massive machine-type (MCT) devices, such as vehicles, sensors, and wearable devices, which brings a large number of application tasks that need to be processed. Additionally, data collected from various devices needs to be executed and processed in a timely, reliable, and efficient manner. In this paper, we introduce the multi-access edge computing (MEC) service into the IoT system such that it can assist the computation offloading and resource allocation for different application tasks. Further, we define a total cost function as the weighted sum of task delay and energy consumption and propose a novel deep reinforcement learning (DRL)-based decision-making algorithm to cooperatively optimize the task offloading and resource allocation. More specifically, we transform the initial problem into a convex optimization problem with the aid of our proposed atom action generation technique (2AGT) and adaptive aggregation parameter update strategy (2APUS). Additionally, the DRL agent will be retrained periodically to adapt to different network environments. Finally, numerical simulation results validate that the proposed algorithm is nearly close to optimal performance compared with some cutting-edge schemes.
Paper
Full text
Deep Reinforcement Learning Aided Computation Offloading and Resource Allocation for IoT
Semantic Scholar · Computer Science · 2020
Abstract
Internet of Things (IoT) expects to incorporate massive machine-type (MCT) devices, such as vehicles, sensors, and wearable devices, which brings a large number of application tasks that need to be processed. Additionally, data collected from various devices needs to be executed and processed in a timely, reliable, and efficient manner. In this paper, we introduce the multi-access edge computing (MEC) service into the IoT system such that it can assist the computation offloading and resource allocation for different application tasks. Further, we define a total cost function as the weighted sum of task delay and energy consumption and propose a novel deep reinforcement learning (DRL)-based decision-making algorithm to cooperatively optimize the task offloading and resource allocation. More specifically, we transform the initial problem into a convex optimization problem with the aid of our proposed atom action generation technique (2AGT) and adaptive aggregation parameter update strategy (2APUS). Additionally, the DRL agent will be retrained periodically to adapt to different network environments. Finally, numerical simulation results validate that the proposed algorithm is nearly close to optimal performance compared with some cutting-edge schemes.