Research On MEC Network Optimization Design For Physical Layer Security And Caching

As wireless communication technologies and smart mobile devices have advanced,people have higher requirements for computing and resource storage.Traditional cloud computing is to compute and store data on the cloud servers which is far away from mobile terminals.But it causes high latency and excessive network traffic during data transfer and resource acquisition.In order to achieve low latency and reduce network transmission traffic,researchers have introduced the concept of mobile edge computing.Although mobile edge computing technology can reduce the latency effectively and improve the user experience in obtaining data,there are many key issues that need to be addressed in mobile edge computing technology in terms of computation and data storage: 1)how to ensure the security of the system when there is an intelligent attacker in the wireless communication network;2)how to ensure the scheduling of system communication and computing resources,when there is an intelligent eavesdropper in the mobile edge computing network;3)how to implement the design of user data caching in the mobile edge computing network.The objective of this paper is to address the above concerns.We firstly consider a communication system model in a cognitive wireless communication network in which there is a secondary intelligent attacker with silent,eavesdropping,jamming and spoofing modes.In the process of sending information from the sender to the receiver,the intelligent attacker will attack for the purposes of eavesdropping,jamming,and spoofing.We consider the static secure communication game between the main link and the attacker,and derive the Nash equilibrium point based on game theory to improve the secrecy of the system.To find the best transmission power level,we devise a reinforcement learning policy.The simulation results show that the optimization method can effectively suppress the attacks of intelligent attacker and assure the security of the system.Secondly,this paper combines physical layer security with mobile edge computing networks based on the above research.We investigate computation offloading decision and communication resources in the presence of intelligent attacker in mobile edge network.We consider a mobile edge network where there are multiple users and an intelligent attacker.Users offload computational tasks to computational nodes because of their limited computational capacity,and an intelligent attacker eavesdrops during the offloading.To ensure the minimization of system latency and energy consumption,we formulate the issue as an optimization problem for offloading ratio,bandwidth and transmit power allocation.To optimize the system,we apply deep reinforcement learning and characterize the optimization issue as a Markov decision process.We create the state space,action space,and reward,as well as a multi-objective optimization technique based on deep reinforcement learning.The safe offloading technique based on a deep reinforcement learning algorithm can achieve convergence and minimize the system latency and energy consumption,according to simulation data.Finally,this paper investigates how to implement user data cache strategy in mobile edge computing networks by combining cache with mobile edge computing networks in data storage.We consider an edge network consists of base station,relays,and users.The base station and the relays have limited cache space.In order to reduce the latency of users in obtaining data and use the limited storage space more efficiently,we design a cache prediction framework by predicting the future request behavior of users through their historical request data and user preferences.We formulate the problem as maximizing cache hit rate and minimizing request latency by analyzing the transmission and computational processes.We use deep neural networks to train and learn users' preferences to improve the hit rate of cache content,and analyze the channels between relays and user to determine cache policies.The proposed framework can improve the cache hit rate and increase the value of latency reduction in circumstances with varying cache spaces and different numbers of users,according to simulation experimental results.