Efficient Collaborative Caching In Cloud Radio Access Network

In recent years,with the explosive growth of mobile data traffic,traditional net-work architecture faces a great deal of challenges,especially the limited backhaul link bandwidth and large energy consumption bring a heavy burden to the network.Cloud Radio Access Network(C-RAN)is a promising architecture for 5G cellular networks to cope with there problems.It can achieve higher capacity,lower energy consumption,as well as cheaper operational cost.Proactive caching strategy in C-RAN can further improve quality of service(QoS)and quality of user experience(QoE),and is a win-win strategy for both operators and users.This paper considers a two-tier collaborative caching architecture in C-RAN,where popular files are cached at both the baseband unit(BBU)and the remote ra-dio heads(RRHs),and the RRHs collaboratively serve the request of mobile users via BBU pool.On this basis,two different cache schemes are designed.1.For network operator,though caching can bring multiple benefits including alleviating the backhaul traffic,reducing the transmission delay,etc.,it also incurs significant additional energy consumption for storage.We introduce MINEC,the min-imum energy consumption collaborative caching problem,where we try to optimize the caching size and caching scheme for each infrastructure such that the total energy cost of both storage and transmission is minimized.For the common routing poli-cy where each mobile user is associated with its nearest RRH,we propose a novel polynomial-time algorithm which solves MINEC exactly.For the shortest path routing policy where each mobile user is associated with multiple RRHs,from the perspec-tive of energy cost,we show that MINEC is NP-hard.We then recast MINEC as a maximization problem targeting at maximizing the energy saving due to caching,and further introduce an approximation algorithm for the maximization problem.The effi-ciency of our proposed caching schemes is demonstrated via trace-driven experiments,which highlights the advantage of applying collaborative caching scheme in BBU and RRHs.2.For users,The more cached files can no doubt result in quick response for content access.But it is impractical to cache all files.Thus,we studied how to de-sign an efficient caching scheme to minimize the latency of user requests for files under the constraint of capacity and bandwidth.This paper models the above prob-lem as a binary integer linear programming problem(BILP)and solves it by decom-posing it into two sub-problems(Content-Placement subproblem and Content-Routing subproblem).Firstly,a greedy-based hierarchical caching strategy is proposed for the Content-Placement subproblem of BBU and RRH with limited cache size.As for Content-Routing subproblem,based on the previous sub-problem,we finally proposed a greedy-based heuristic algorithm under the constraint of bandwidth to determine the cache placement scheme and routing scheme.Simulation results show that our algo-rithm can effectively reduce 26%users' latency compared to the classic cache algorith-m,and indicate that different BBU/RRH cache size settings will affect the performance of users',average delay.