The Research Of Blockchain-based Intelligent Caching Strategy In Edge Computing Network

The evolution of the sixth generation mobile communication(6G)will put higher requirements for network performance such as transmission rate,end-to-end delay,reliability and spectral efficiency.As one of the solutions for improving network performance,edge computing places computation and storage resource to the edge network to meet the needs of computation-intensive and delay-sensitive applications.Edge caching is an important part of the edge computing network.It can be used to cache content data to reduce the delay of acquiring content,or cache the service program to support computation offloading.Compared to traditional cloud storage,edge caching is more distributed and flexible,facing greater challenges in adaptive resource allocation and network security.Therefore,how to assign caching resource in an intelligent way while ensure the credibility of caching data in the distributed network is a question which is worth studying.As an emerging distributed system,blockchain is widely concerned in the academic and industrial fields for its distributed,data-traceable and non-tampered storage.By integrating blockchain to the edge computing network,different types of resources can be managed and allocated in a self-organized way,and then a more efficient,trusted and intelligent edge caching system can be established.However,edge nodes are often resource-limited,and the normal operation of the traditional blockchain depends on a large amount of computation and storage resources.Simply mapping edge nodes to the blockchain network would bring large overhead to the network because of data storage and synchronization.Therefore,this thesis studies the blockchain-based edge content caching and edge service caching.Meanwhile,this thesis designs lightweight and high efficiency consensus mechanism for edge computing network,and use intelligent methods to achieve safe and efficient edge caching.The main work of this thesis is summarized as follow:This thesis designs a blockchain-based collaborative caching framework and a trustworthy content sharing model in the edge content caching scenario,where the storage resource of both user equipment and edge servers are fully utilized.In order to select the cached content intelligently under the storage constraint,this thesis proposes a predictive model for content popularity based on linear regression,which enables the node to proactive cache according to the trend of content popularity.The credit and tokens in the blockchain can encourage nodes to cache and transmit more content in honest behavior,and untrusted nodes will pay for their malicious actions such as tampering or deleting cached data.Since each node chooses strategy independently to maximize its benefits in an environment of mutual influence,a non-cooperative game model is designed to study the caching behavior among edge nodes.The existence of Nash equilibrium(NE)is proved in this game,so the collaborative edge caching allocation algorithm is proposed to choose the optimal caching strategy for collaborative devices according to content popularity,node credit and other nodes’ caching decisions.The trusted nodes will get the maximum caching rewards under the storage space limitation.Simulation results show that the system can enhance trust in the edge collaboration caching as well as improve the caching efficiency.This thesis studies the intelligent optimization strategy of service caching and computation offloading based on blockchain in the edge service caching scenario,where service caching is used to support the computation offloading of user equipment.In the edge computing network,users need to upload the corresponding service program when performing computation offloading.With the credibility assessment mechanism and proof-of-benefit consensus in the blockchain,the edge server can ensure the credibility of the cached service and reduce the overhead of data synchronization in the blockchain network.In order to allocate caching and computation resources intelligently,the overall delay and service credibility are jointly optimized.This thesis proposes a approach which bases on deep reinforcement learning to make offloading decision,and carry out service caching according to both credibility and offloading decisions of multiple users.Simulation results show that compared with existing schemes,the proposed system can effectively reduce the overall delay as well as improve the credibility of the network.