Research On Brain Storm Optimization-based Hierarchical Cooperative Caching In Fog Radio Access Network

In recent years,global mobile data traffic has experienced tremendous and continuous growth with the proliferation of mobile devices and multimedia services.The increasing data traffic imposes enormous pressure on backhaul links with limited capacity and long time delay,which can potentially result in unprecedented traffic congestion,especially during peak hours.Fog radio access network(F-RAN)has been recognized as a promising architecture to mitigate traffic congestion and reduce content request delay by placing popular contents at fog access points(F-APs)which are closer to users.However,the contradiction between limited storage capacity and diverse user requests poses challenges to the implementation of effective caching policy.To solve these challenges,the hierarchical cooperative caching problem in F-RAN is investigated in this paper.Firstly,the brain storm optimization(BSO)-based hierarchical cooperative caching policy is studied.To minimize the content request delay,we formulate the hierarchical cooperative caching optimization problem based on local content popularity and find the optimal caching policy,where both horizontal cooperation among fog access points(F-APs)and vertical cooperation between the cloud server and F-APs are jointly considered.Considering the non-deterministic polynomial hard(NP-hard)property of this problem,we propose a brain storm optimization(BSO)approach which utilizes the penalty-based fitness function in individuals evaluation to meet the storage capacity constraint and the chromosome representation in new individuals generation to meet the integer constraint,respectively.Moreover,to reduce the computational complexity,we propose to implement the convergent operation in the objective space rather than solution space via individuals classification.The simulation results show that the proposed BSO-based hierarchical cooperative caching policy can effectively reduce the average content request delay.Secondly,the BSO-based clustered hierarchical cooperative caching policy is studied.According to local content popularity similarity and geographical location similarity of F-APs,a sample set is constructed.The distance measurement method between points in the set is defined and density-based spatial clustering of applications with noise(DBSCAN)is utilized to form F-AP clusters.In order to improve content diversity in the same cluster,we propose to eliminate the content redundancy and ensure that one content will not be repeatedly cached in the same cluster.Based on the new F-AP clusters,a hierarchical cooperative caching optimization problem is formulated to minimize content request delay with three constraints: the limited storage capacity of F-APs and the cloud server,the integer characteristics of caching decision variables and the elimination of the content redundancy in one cluster.The BSO approach is utilized to solve this optimization problem.We propose the individual modification operation to satisfy the non-redundant constraint in one F-AP cluster.The simulation results show that the proposed BSO-based clustered hierarchical cooperative caching policy can effectively reduce the average content request delay.Finally,the improved distributed BSO-based clustered hierarchical cooperative caching policy is studied.The Opposition-based Learning(OBL)operation is utilized to improve the initial solution space,which can broaden the search space and enhance the initial searching performance of the algorithm.Moreover,the dynamic penalty function is utilized to improve the individual evaluation operation.At the earlier stage of the algorithm searching procedure,a smaller penalty value will make the individuals try to diverge in the search space,increasing the diversity of the population.At the later stage of the algorithm searching procedure,a larger penalty value will enhance the convergence of the algorithm to the global optimum.In addition,considering the computing resources in F-APs,we propose the parallelization of the BSO approach to assign the computing tasks of the cloud server to F-APs.The single BSO population is divided into several sub-populations,and each F-AP cluster undertakes the evolution task of the sub-populations.The designed interactive operation improves the search performance of the algorithm by sharing the evolutionary results among the sub-populations.The theoretical analysis and simulation results show that the improved distributed BSO-based clustered hierarchical cooperative caching policy converges to the global optimum and effectively reduces the average content request delay.