针对毫米波(mmWave,mmW)移动边缘计算(Mobile Edge Computing,MEC)系统在高密度城区环境下难以获得持续稳定、高质量、高安全性的边缘缓存问题,提出一种基于联邦学习(Federated Learning,FL)的可重构智能超表面(Reconfigurable Intelligence Surface,RIS)辅助MEC系统边缘缓存算法.该算法首先通过RIS构建高质量的反射信道,改善了高密度城区随机阻塞环境带来的影响;其次基于联邦学习构建一个高效的全局模型,避免了用户数据的隐私泄露;最后,通过引入热门内容预测算法和Dueling Double Deep Q-Network(D3QN)算法,以获得最优的缓存位置,从而提升了MEC系统的边缘缓存效率.仿真实验验证了所提算法的有效性.
Edge Caching Scheme for RIS-Assisted MEC Systems Based on FL
Addressing the challenges of achieving consistent,stable,high-quality,and secure edge caching in millimeter-wave(mmWave)Mobile Edge Computing(MEC)systems in high-density urban environments,an edge caching algorithm for Reconfigurable Intelligence Surface(RIS)-assisted MEC systems based on Federated Learning(FL)was proposed.The algorithm first constructed high-quality reflection channels through RIS to mitigate the impacts of random blockages in high-density urban environments.Secondly,it build an efficient global model based on FL to prevent privacy leakage of user data.Finally,by introducing a popular content prediction algorithm and the Dueling Double Deep Q-Network(D3QN)algorithm,optimal cache placement was achieved,thereby enhancing the edge caching efficiency of the MEC system.Simulation experiments verified the effectiveness of the proposed algorithm.
万方数据
维普
