Joint Computation Offloading and Task Caching Strategy for MEC-Enabled IIoT

Abstract

Industrial Internet of Things (IIoT), as a key link in the transformation of traditional manufacturing to digitalization, can be paired with Multi-access Edge Computing (MEC) technology to satisfy the low-latency environment required by industry. Nonetheless, the system contends with uncertain environmental factors such as dynamic changes in channel state and random task generation. Motivated by these, this paper designs an intelligent offloading and task caching strategy to reduce the overall execution latency of tasks. The interaction process within system is modeled as an Markov Decision Process (MDP), and we introduce a low-latency scheduling strategy leveraging Deep Reinforcement Learning (DRL), termed DDPG-LL. Besides, the proposed strategy is tailored for optimizing the task queue of the MEC server. By considering factors such as priority, waiting time, and completion expectations, queue adjustments are dynamically made at each time slot. Simulation results demonstrate that the proposed strategy achieves rapid and stable convergence, and effectively reduces the completion latency of tasks compared to the baseline strategies.