Research On Coded Caching In Fog Radio Access Networks

High-rate applications such as video streaming have become the mainstream of file requests,and this has brought unprecedented pressure to traditional wireless networks.Due to the constrained fronthaul capacity and centralized baseband unit pool,traditional centralized cloud radio access networks(C-RANs)have been unable to meet the high-quality and high-real-time requests during the peak hours of the networks.In order to meet the increasing demand for communications,the application of fog radio access networks(F-RANs)in edge caching and edge computing has recently attracted widespread attention.The fog access points(F-APs)in F-RANs are close to the users,which can improve network congestion and quality of service by using edge caching resources.However,due to the constrained caching resources and the randomness of user requests,edge caching resources are often not sufficiently utilized.By elaborately designing the cache contents and content delivery via coded caching,the conflict above can be effectively alleviated.Nevertheless,different coded schemes,placement schemes and delivery schemes lead to huge differences in the performance of fronthaul load.In this thesis,we will systematically study the coding schemes,coded placement schemes,and coded delivery schemes to solve the above problems.Firstly,a joint redundant maximum distance separable(MDS)codes and cluster cooperation based coded caching policy is studied.A MDS code is first adopted as the choice for the coded packet.The trade-off between the redundancy and reliability of the MDS codes can simplify the placement and delivery schemes.Then,based on a user-centric service model,a redundant MDS code based coded placement scheme is proposed.The proposed placement scheme can reduce the complexity of the placement phase and provide multicast opportunities in the delivery phase.Based on the proposed placement scheme,a cluster cooperation based coded delivery scheme is then proposed,which can use the cache contents of any two clusters to construct multicast messages.In addition,a joint MDS codes and cluster cooperation based coded caching policy is proposed.This policy can use redundant coded packets to create multicast messages for consistent requests,and use the cluster cooperation based coded delivery scheme to create multicast messages for both consistent and inconsistent requests.Finally,the performance of the policy is theoretically analyzed,and the expression of the fronthaul load is derived,and the relationship between the fronthaul load and the number of redundant coded packets and the number of file fragments and other related parameters are explored.Secondly,a joint MDS codes and weighted graph based coded caching policy is investigated.Based on the above coded placement scheme,the cluster cooperation based coded delivery scheme is generalized.Multiple clusters in a cluster group can be used to create multicast messages.By viewing the clusters as vertices and the multicast opportunities between two clusters as edges,we establish the relationshion between the design of delivery scheme and the graph theory.The coded delivery problem can then be transformed into an integer programming problem.Furthermore,a weighted graph based coded delivery scheme is proposed,which can further explore the multicast opportunities among clusters.Moreover,a joint MDS codes and weighted graph based coded caching policy is proposed,and we analyze the performance of the proposed policy by deriving the upper and lower bounds of fronthaul load.Furthermore,we show through theoretical analysis that the performance gap between our proposed policy and its lower bound is extremely small.Simulation results also show the superior performance of our proposed policy in comparison with the other traditional policies.Finally,a hypergraph strong coloring based coded placement scheme and cluster match based coded delivery scheme are investigated.By extending the system model to the more general situation,where a certain number of F-APs can be reused in different clusters and different F-APs in different clusters are allowed to cache same coded packets,the utilization of all F-APs can be improved and the MDS coding loss can be reduced.In addition,by viewing the F-APs as vertices and the clusters as hyperedges,the coded placement problem is equivalent to a hypergraph strong coloring problem,and a heuristic cache placement scheme based on a hypergraph strong coloring is proposed.Moreover,for a given placement scheme,a cluster match based coded delivery scheme is proposed to construct multicast messages during the delivery phase.The superior performance of the proposed coded delivery scheme in reducing the fronthaul load is verified by simulation results.