Deep Reinforcement Learning-based Edge Caching for Industrial Control Applications

To enhance the quality of service for field users in industrial control applications, a suitable caching strategy at edge servers is essential. This paper proposes a cache replacement strategy based on deep reinforcement learning. Status space, action space and reward function are defined considering the varying real-time requirements of the application files. The performance of the proposed algorithm is compared with baseline algorithms using user requests with dynamic popularities. The experimental results demonstrate that the proposed algorithm can effectively enhance the hit rate of control files while maintaining the overall cache hit rate, without sacrificing performance.