Research On Learning-based Optimization For Mobile Cooperative Edge Caching

With the rapid development of mobile Internet,mobile network traffic shows explosive spikes.Mobile users are increasingly demanding services for various content files.Therefore,in order to meet these challenges,it is very necessary to introduce new technologies to the next generation mobile network architecture to cope with the current situation of traffic surge.In recent years,deploying caches at the edge of mobile networks is seen as a promising technology to improve the reusability of content within the network by caching popular content on edge devices such as base stations,avoiding duplicate content consuming large amounts of network resources.However,the overall optimization of traffic optimization for mobile edge caching is lack of comprehensiveness and adaptability.At the same time,many related researches also lack an efficient collaborative caching mechanism,resulting in low cache performance.In this paper,we study the edge caching optimization of mobile networks.Traditional cache solutions lack the ability of learning iteration and adaptive capabilities in time dimension.We model the content caching and request routing problem as a Markov Decision Process(MDP).By virtue of Deep Reinforcement Learning(DRL)with respect to solving complicated control problems,we propose a mobile edge cooperative caching strategy based on learning mechanism.Particularly,we aim at minimizing the long-term average content fetching delay of mobile users without requiring any priori knowledge of content popularity distribution to evaluate and track changes in content popularity in a timely manner.At the same time,this paper proposes a distributed collaborative training mechanism based on federated learning,which is used to quickly train the model while protecting data privacy.Trace-driven simulation results show that our proposed framework outperforms some existing caching algorithms,including Least Recently Used(LRU),Least Frequently Used(LFU)and First-In First-Out(FIFO)caching strategies by 7%,11% and 9% improvements,respectively.Besides,our proposed work is further shown that only average 4%performance loss exists compared to an omniscient oracle algorithm.