Research On D2D Content Distribution Strategy With Feature Perception

With the explosive increase in user demand for low-latency services and the exponential growth in mobile device access,resource on the cellular side is becoming tighter.How to deal with the huge data brought by these services has become an urgent problem for mobile communications.In order to solve these problems,the 3rd Generation Mobile Communication Partnership Project(The Third Generation Partner Project,3GPP)proposes a direct device to device(D2D)technology to reduce base station load pressure and improve spectrum utilization.At the same time,the European Telecommunication Standardization Association also proposed the concept of edge computing,sinking some functions of the core network to the edge network to reduce transmission delay.Combining caching technology in edge computing can promote sharing among users by D2 D.First,this thesis introduces the related concepts of D2 D communication and typical application scenarios,and analyzes and summarizes the current research hotspots.In particular,the cache selection strategies in D2 D combined with edge cache are summarized.Further,the resource allocation strategy in the content transmission process is analyzed.Then,a content caching strategy with feature awareness is proposed for the cache content selection problem.For the repeated transmission of a large amount of content in the local network,the content is cached in the local node and users are allowed to share the content through D2 D communication.Specifically,the edge network is divided into a macro base station,a micro base station,and a user equipment three-level cache node,where the user equipment caches used content by default.Content popular value is divided into current value and future value,current value is related to user characteristics and content characteristics,and future value is related to user social relationship.The micro base station chooses to cache content with higher user value in the coverage area,and the macro base station filters common parts from each micro base station for caching.Numerical results show that the proposed strategy can effectively reduce user delay.Furthermore,a resource allocation strategy with feature awareness is proposed for cochannel interference to cellular users during content transmission.First,the resource allocation problem is decomposed into two sub-problems: physical resource block allocation and power allocation.Second,quantify business characteristics and user characteristics related to resource allocation.In order to promote users to share content,these characteristics determine the user's priority in the resource allocation process.Then,based on the allocation of Q-learning,a forgetting gate in long-term and short-term memory networks is introduced to solve the problem that the Q-learning method may fall into a local optimum.Simulation results show that the proposed strategy has higher transmission rate and spectrum utilization.