Optimal Design Of Cache Placement And Content Delivery In Fog Radio Access Networks

5G is committed to providing users with ultra-high definition,ultra-low latency video traffic transmission services,while supporting massive connections to achieve the ultimate goal of all things connected.In the era of mobile Internet,mobile data traffic is exploding.Traditional wireless access networks cannot fully utilize the scarce spectrum resources,so they are no longer suitable for the upcoming 5G communication system.Cloud radio access network(Cloud RAN)and fog radio access network(Fog RAN)are emerging network architectures in recent years.In Cloud RANs,the remote radio unit(RRU)is responsible for radio frequency,and the central baseband unit pool(BBU Pool)integrates computing resources for data processing and resource scheduling.Data exchange between the BBU Pool and the RRU is implemented by Fronthaul.Based on the Cloud RAN,Fog RAN enables the RRU to have caching and computing functions.Such RRU is simply referred to as the edge node(EN).In Fog RANs,joint scheduling of multiple ENs through the BBU Pool can greatly improve the access layer spectrum efficiency.However,the joint transmission between multiple ENs need to share the same data,which leads to great burden on Frontaul.Therefore,the Frontaul link overhead and access layer performance are contradictory.Since the popular video traffic is repeatedly requested in a short time,pre-caching it to the EN can greatly reduce the Fronthaul load and the access layer transmission delay.So the cache can effectively alleviate the contradiction between the Fronthaul link and the access layer.However,compared with the huge video traffic,the cache resources are limited.Therefore,how to design the proper caching strategy to maximize the cache utilization is a hot topic in recent years.The introduction of caching will also change the content delivery mode.Designing the suitable transmission strategy based on the cache state is another research hotspot.This thesis focuses on cache placement and content delivery issues in Fog RANs.The whole work is divided into two parts.The first part mainly explores how to optimally allocate the cache space to multiple files through partition-based coded caching,thus alleviating the contradiction between Frontaul link overhead and access layer performance.The caching strategy design consists of the transmission phase and the cache phase.In the transmission phase,the resource allocation mainly consider the instantaneous information such as the user request and the channel state at a certain moment.Then in the cache phase all possible transmission scenarios are comprehensively considered.The expectation of the optimal objective value for the delivery phase is considered as the performance criteria for cache allocation.By analyzing the nature of the two-stage problem,we propose a low-complexity greedy algorithm to obtain the optimal cache allocation result.The simulation results show that the proposed caching strategy can achieve the best balance between Frontaul link overhead and access layer performance.In addition,the simulation results also show that the optimal caching strategy will change with different Frontaul cost measurement functions,and the partition-based caching is always better than caching without partition.The proposed caching strategy can apply to any convex Frontaul cost measurement function,so its versatility is very high.The second part mainly explores how to transfer content according to the cache state in Fog RANs with multiple multi-antenna ENs and multiple target users.The cache state is based on the partition-based caching.We make each user apply the successive interference cancellation decoder(SIC Decoder)for multiple subfile data streams reception and propose a method for determining the reception order based on local cache and channel state.The transmission strategy design includes EN clustering and multi-antenna precoding,and the corresponding optimization problem is a mixed integer nonlinear programming.An algorithm combining the convex-concave procedure(CCP)and alternating direction method of multipliers(ADMM)is proposed to solve this problem.The simulation results show that the proposed algorithm is more efficient than the CCP with interior-point methods.As the network scale becomes larger,the algorithm complexity grows slowly,and the distributed nature of ADMM is more applicable to Fog RANs.In addition,the simulation results also show that the decoding order has a great impact on performance when using SIC technology,and the method for determining decoding order proposed in this thesis can make full use of local cache resources,which has great advantages when Fronthaul resources are limited.Finally,the simulation further proves that in the complex Fog RAN transmission scenario,the partition-based caching-transmission is still better than caching-transmission without partition.