Cache Mechanism With Privacy Protection Capability In F-RAN

As 5G enters the commercial stage,emerging applications such as virtual reality,augmented reality,Internet of vehicles,and holographic imaging are developing rapidly.The demand for content services is increasing.Facing the rapid growth of mobile data traffic,in the Fog Radio Access Network(F-RAN)is expected to improve the quality of network services by combining artificial intelligence and edge caching technology.However,data privacy issues brought about by artificial intelligence technology are becoming more and more serious.In order to tackle with the data privacy problem that artificial intelligence in F-RAN faces when optimizing cache resources,this paper proposes a cache mechanism with privacy protection capabilities in F-RAN.First,in the cache resource optimization scenario,the delay optimization problem and the privacy protection of terminal data were researched.Second,in the joint optimization scenario of cache resource and power allocation,the joint optimization and the privacy protection of terminal data were studied.The main ideas and innovations of the thesis are summarized as follows:1.For the delay optimization problem of F-RAN,this paper adopts an active caching strategy based on deep reinforcement learning to enhance the edge content service capabilities.Based on the F-RAN distributed network architecture,content services can adaptively satisfy user requests for content through the F-AP mode at the edge of the network or the C-RAN method at the cloud layer.In addition,to protect the privacy of terminal data,this paper proposes a federated learning framework suitable for the optimization of cache resources in F-RAN.The federated learning mechanism is used to help achieve flexible trilateral collaboration between user equipment,fog wireless access nodes,and cloud content servers.Active cache placement can be optimized under the condition that user data is stored locally.Finally,the system simulation proves that the designed mechanism is effective in reducing the average delay of content access and improving the hit rate.Compared with the traditional caching mechanism,its performance is better than the traditional caching algorithm.2.As to the time delay and power allocation of F-RAN,this paper proposes a deep reinforcement learning method to jointly optimize the cache resource and power allocation.First,consider terminal privacy issues in the optimization process,and the federal learning framework is used for terminal data privacy protection.Then we consider the collusion between the aggregation node and multiple users and use the secret sharing protocol and key agreement algorithm to protect the privacy of the neural network model.Finally,in order to solve the problem of model integrity,homomorphic hash function and bilinear mapping are used to verify the integrity of the model.The simulation results prove that the designed joint optimization strategy can effectively improve the network performance.The cryptographic technology used to achieve privacy protection has low overhead.In summary,this article focuses on the optimization of cache resources in the fog wireless network and the privacy issues arising from the use of artificial intelligence methods.It is a useful supplement to the fog wireless network resource optimization and privacy protection methods.At the same time,it also provides new ideas for the study of privacy issues in resource optimization in other wireless networks.