Research Of Data Cache Strategies For Mobile Edge Computing

With the explosive growth of data traffic in wireless communication networks,relying on center resource pool to serve massive mobile terminal equipments shows multiple disadvantages.The traditional cloud radio access network presents significant bottleneck in terms of service delay.Computing through edge nodes in the network proves to be a promising way to relieve the front-haul overload and cope with the demand for massive data.Cache-enabled wireless networks such as small-cell cellular wireless communication network and device-to-device communication network can effectively distribute the data load of core network,reduce service delay as well as improve quality of experience.However,subject to the limited cache capacity of edge nodes,how to design an effective edge cache strategy is a significant problem to be solved.This thesis focuses on cache strategies for small-cell cellular networks,and proposes three cache strategies including optimal non-cooperative cache strategy towards both cache hit ratio and transfer delay,optimal cooperative cache strategy towards maximal cache hit ratio and optimal cooperative cache strategy towards minimal transfer delay.Main work of this thesis is summarized as follows.Firstly,in order to improve the average cache hit ratio and reduce the average file transmission delay in non-cooperative cache scenario via designing cache placements of small base stations,non-cooperative cache strategies towards maximal average cache hit ratio and minimal average file transfer delay are respectively proposed.Taking advantage of the cache independence among small base stations in non-cooperative cache scenario,the optimization problem of maximizing the network's average cache hit ratio and minimizing the network's average file transfer delay are respectively decomposed into the sum of sub-problems of maximizing the average cache hit ratio at each small base station and sub-problems of minimizing the average file transfer delay at each small base station.Optimal solutions of above two optimization problems are given in closed form via cache hit ratio analysis and calculation.Moreover,the cache strategy maximizing the average cache hit ratio and the cache strategy minimizing the average file transfer delay are proved to be consistent in non-cooperative cache scenario.Simulation results show that the proposed non-cooperative cache strategy can significantly improve the average cache hit ratio and reduce the average file transfer delay of the network.Secondly,aiming at the problem of improving the average cache hit ratio of the network by designing cache placements of small base stations in cooperative cache scenario,a cooperative cache strategy towards maximal average cache hit ratio is proposed.The average cache hit ratio of the network is analyzed on the basis of file request probabilities and the cooperative transmission relationship among small base stations,and by decomposing the original optimization problem into the sum of several independent sub-problems,the optimal solution is derived in closed form.Moreover,we proposed to model the network as a two-dimensional directed topology according to the cooperative transmission relationship among small base stations.By precisely designing the cooperative relationship matrix,cache placement matrix and request probability matrix,the matrix expression of the original optimization problem is established,and convex optimization tools are utilized to solve the problem.Simulation results show that the proposed optimal cooperative cache placements towards maximal cache hit ratio can achieve higher cache hit ratio performance than present cache strategies such as random cache and most popular cache,and significantly reduce the front-haul overload at the same time.Thirdly,aiming at the problem of reducing the average file transfer delay of the network by designing cache placements of small base stations in cooperative cache scenario,a cooperative cache strategy towards minimal average file transfer delay is proposed.The cooperative transmission model among small base stations is established according to the relative size of the unit file transfer delay between any pair of small base stations and the unit file transfer delay from the macro base station to small base stations.In order to solve the problem that it is hard to derive the delay-optimal cache placements in closed form,we propose to transform the original cooperative cache strategy optimization problem into the combination of non-cooperative cache strategy optimization problem and limited times of cache substitution,and derive the nearoptimal cooperative cache placements towards minimal transfer delay via a two-stage optimization algorithm.Moreover,we propose to model the cooperative network as a two-dimensional directed topology.The matrix expression of the original optimization problem is established by precisely designing the transfer delay matrix,cache placements matrix and request probability matrix,then the problem is solved by utilizing convex optimization tools.Simulation results show that the proposed cooperative cache placements towards minimal average file transfer delay can achieve better average file transfer delay performance and front-load reduction ratio performance than random cache and most popular cache strategies.