Research On Content Placement Of Cache-enabled Edge Network

Mobile data traffic keeps increasing in recent years due to the growing popularity of mobile devices,videos and various applications.For the traditional content delivery network,all contents are stored on remote servers and they can be sent to users via backhaul network.The increasing data demands of mobile users may result in heavy burden on backhaul network and degrade quality of service(Qo S).Compromised by the ever increasing amount of mobile data traffic generated by mobile devices,edge servers are expected to provide caching resources for users so as to enable low-latency content delivery and release the burden on backhaul network.Edge caching is considered as a promising technology to provide low-latency content delivery in the future 5G network,where the dense heterogeneous base stations(BSs),including macro BSs,femto BSs and pico BSs,can provide abundant caching resources.An important issue in edge caching network is how to design content placement strategy for the edge servers so as to improve the hit probability for a specific user.The hit probability here refers to the probability that a user can successfully get its requested content from edge servers rather than remote servers.The hit probability of the caching network is the same with the offloading rate of backhaul links.For the wireless edge caching network,it is difficult to decide the caching strategy because of the unknown topology of networks,the large-scale of caching scenario and the mobility of the users.Probabilistic caching strategy,where contents are cached in edge servers according to the same distribution is proposed for large scale wireless caching network.Researchers can optimize the performance of caching network via seeking optimal caching probability.Based on the probabilistic caching,we study the performance of probabilistic caching in two-tier heterogeneous caching network,propose a spatially cooperative caching strategy and study the performance of physical layer security(PLS)in caching network.The main contributions of this thesis are summarized as follows.Firstly,we consider optimal probabilistic caching with different sizes of contents in a twotier heterogeneous caching network.We use Lagrangian multiplier method to solve this convex problem and derive the closed-form solution of Lagrangian multipliers as well as the optimal caching probabilities rather than resorting to the bisection method which is superfrequently used in most available works.We proposes caching probability conversion(CPC)algorithm to derive the accurate optimal caching probabilities.Compared with the common method in previous works,the optimal caching probabilities derived by CPC algorithm can exactly satisfy the caching capacity constraint of each caching devices.This algorithm can avoid the waste or overflow of caching resources theoretically.We extend the network model with the consideration of limited serving capacity constraint.In this case,a caching device can only send one content to the limited number of users simultaneously.We derive the probability that a user will be served by an edge server or a caching helper,analyze the average hit probability and solve the non-convex problem by difference of convex(DC)programming.Secondly,We propose spatially cooperative caching strategy for a two-tier heterogeneous network based on the probabilistic caching strategy.In this strategy,we divide the caching helpers into two kinds: the covered helpers and uncovered helpers.For the covered helpers,the cannot cache the contents which have been cached by its nearest edge server.The proposed strategy can efficiently reduce the duplicate caching in the helpers,improve caching resource utilization in caching helpers and improve the performance of the caching network.We develop an analysis framework for the proposed spatially cooperative caching strategy by using tools from stochastic geometry.Performance metrics are analyzed from the perspective of hit probability considering the effects of BS density,helper density,communication ranges,popularity distribution,etc.The simulation results are identical with the analytical ones which means our theoretical model can efficiently characterize the performance from hit probability of the proposed caching strategy in a two-tier heterogeneous network.We maximize the hit probability of the two-tier heterogeneous network by optimizing the caching probabilities of contents combinations on edge servers and caching helpers.Particularly,we derive the closed-form solution of the optimization problem for unit cache size.The outperformance is observed with the optimal caching probability derived in this paper compared with the existing caching strategies.Thirdly,we consider the physical layer security issue on edge caching network.we consider physical layer security in large-scale probabilistic caching and focus on maximizing secure content delivery probability problem in large-scale edge caching network.We analyze secure delivery probability of caching network in the presence of randomly distributed eavesdroppers,jointly optimize the redundant rate and caching probabilities by our proposed alternative optimization algorithm and make comparisons with existing caching strategies.