Research On Edge Caching Strategy With Social Attribute

With the rapid growth of mobile services,social applications and resource-intensive applications,mobile data traffic is exploding exponentially.As a promising architecture,mobile edge networks utilize the low-cost resources for computation and caching which are located on the edge of network.As a result,the transmission delay can be reduced and the load of the backhaul link can be alleviated effectively.Edge caching is regarded as an effective solution to solve the problem of large-scale content interaction in the future network.Besides,the combination of the device-to-device(D2D)communication and social attributes can guarantee the performance of short-distance communication and improve the Quality of Service(Qo S)effectively.Therefore,the D2 D enabled edge caching scheme based on social attributes has been widely studied by the academics and industries.Firstly,the related concepts,characteristics and typical architectures of mobile edge networks are introduced.Then the current research is analyzed.The scenarios and the methods of mobile edge caching are described briefly.The existing mobile edge caching strategies are described,and the edge caching strategies based on the social attributes are analyzed in depth.Secondly,a socially aware caching strategy in D2 D enabled fog radio access networks(F-RAN)is proposed.Based on the content preferences of fog access point(F-AP),the projective adaptive resonance theory neural network(PART NN)is used to design the initial construction of fog communities.Then the construction of fog communities is updated based on the purpose of the migration.The data is cached collaboratively by the F-APs which belong to tha same fog community.Considering the social attributes and content preferences of users,the proactive cache rate of user is defined by quantifying the forwarding willingness and caching willingness,and then the selection scheme of center user is proposed.Due to the selfishness and the limited cache resources of users,the cache scheme of the center user is proposed based on ant colony algorithm.Numerical results show that the proposed strategies can improve the cache hit ratio effectively and have a lower redundant request rate and transmission delay simultaneously.Thirdly,a D2 D caching strategy is proposed based on social attributes and the Stackelberg game.The cost strategy for incentive of the base stations and the resource strategy of pre-cache users are modeled as a multi-master and multi-slave Stackelberg game.Based on the content popularity,D2D-incented percentage and mobility,the utility functions are quantified and the existence of Nash equilibrium is proved.The best-response dynamic model is used to obtain the optimal solution.Subsequently,a D2 D caching strategy based on Q-learning is proposed.After considering the content popularity and social attributes,the average satisfaction degree of users within the community is quantified by analyzing the community's cache hit ratio and the average influence of pre-cache users.The reward function of Q-learning is defined based on the average satisfaction degree of users within the community.Numerical results show that the proposed strategy can effectively improve the cache hit ratio and satisfaction degrees of user and community.Finally,the contributions of this thesis are summarized and the future researches are prospected.