A3C-Based Dynamic Content Replacement Strategy for Mobile Edge Networks
Guangbin Xiao, Junmin Zhu, Qian Liu · 2023
Facing the challenge of escalating network data traffic due to the proliferation of mobile devices, edge caching emerges as a pivotal solution to alleviate network performance degradation. This paper introduces a dynamic content replacement strategy employing Deep Reinforcement Learning (DRL) to navigate the fluctuating trends in content popularity adeptly. We formulate an optimization objective focused on minimizing the content retrieval delay, which is conceptualized through initial modeling and further refined via the Markov Decision Process (MDP) framework. The Asynchronous Advantage Actor Critic (A3C) algorithm is then applied to decipher this optimization challenge, leading to the derivation of an optimal cache replacement strategy. Through simulation, our approach is evidenced to enhance network performance by reducing delay and augmenting cache hit rates, marking a significant improvement over conventional caching methodologies.