Abstract
With the development of the industrial internet of things, an increasing number of intelligent terminal devices are being deployed in mining operations. However, due to the surge in network traffic and the limited availability of computational resources, these terminal devices face challenges in meeting high-performance requirements such as low transmission latency and low energy consumption. To address this issue, this paper proposes a method that combines partial offloading with collaborative mobile edge computing (MEC). This approach leverages device-to-device communication to partition computational tasks into multiple subtasks, offloading some of them to collaborative devices or MEC servers for execution. This not only alleviates the computational burden on MEC servers but also makes full use of the idle computing resources of terminal devices, thereby enhancing resource utilization efficiency. Given the limited computational capacity of terminal devices, this paper optimizes the offloading decision-making process between terminal devices and MEC servers. By introducing weighted coefficients for latency and energy consumption, the proposed method ensures that task completion latency does not exceed a predefined threshold while minimizing the overall system cost. The problem is formulated as a multi-objective optimization problem, which is solved using a two-layer alternating optimization framework. In the upper layer, an improved genetic algorithm (IGA) based on heuristic rules is employed to generate an offloading decision population, while the lower layer utilizes the deep deterministic policy gradient (DDPG) algorithm to optimize the offloading strategy and the weighted coefficients for latency and energy consumption. To evaluate the effectiveness of the proposed method, we compare it with five baseline algorithms: the improved grey wolf optimizer metaheuristic algorithm, the traditional genetic algorithm, the binary offloading decision mechanism, the partial non-cooperative mechanism, and the fully local execution mechanism. Simulation results demonstrate that the proposed IGA-DDPG algorithm achieves significant improvements over these baseline methods. Specifically, under various experimental scenarios, IGA-DDPG reduces latency by an average of 24.5%, decreases energy consumption by 26.3%, and lowers overall system cost by 44.6%. Moreover, the algorithm consistently ensures a 100% task completion rate under different system configurations.