Performance Analysis Of Cellular Network Caching With Stochastic Geometry

With the development of communication technology and mobile internet service,mobile data traffic exhibits exponential growth,which brings great pressure to the mobile network,results in the waste of bandwidth,energy and server resources,and greatly reduces the quality of user experience.The cellular network caching is proposed to cope with the challenge brought by the growth of data traffic.The core idea of the cellular network caching is to exchange the cache overhead for the bandwidth gain.The content is stored in the base stations(BSs)or even the user equipments(UEs)during the off-peak time.By shortening the distance between the user and the content,the cellular network caching can alleviate the traffic load and decrease the delay.The topic of this thesis is from National Natural Science Foundations of China.This thesis explores the theoretical origin of the cache technology to improve the network performance in cellular network and studies the key factors that affect the performance gain of the cellular network caching,aiming to provide theoretical basis for the deployment and optimization of different forms of cellular network caching.Revolving around two stages of the cache technology,namely the cache placement and the content delivery,this thesis conducts theoretical performance analysis of the homogeneous cellular network caching,heterogeneous cellular network(Hetnet)caching and Unmanned Aerial Vehicle(UAV)assisted cellular network caching.In the corresponding network,this thesis proposes the caching placement strategy and content delivery strategy.Based on the strategies,this thesis derives theoretical expressions of the network performance with the aid of the stochastic geometry theory and analyzes the effect of different network parameters on the network performance.Further,this thesis optimizes the network deployment and the caching strategy to improve the network performance.The concrete content is summarized as follows:First of all,the network performance of the homogeneous cellular network caching is studied with the stochastic geometry.A BSs deployment algorithm is proposed to meet the requirement of the cache coverage probability given the impact of the traffic load distribution on the network performance.To solve the problem of the content delivery constrained by the limited backhaul capacity,a connection-centric delivery strategy considering backhaul capacity allocation is proposed,and the expressions of the successful content delivery probability(SCDP)and the energy efficiency are derived.In the unlimited backhaul scenario,the expression of the cache coverage probability based on the most popular caching placement strategy and the connection-centric delivery strategy is derived considering the characteristics of the traffic load distribution.The optimal BS density is obtained to satisfy the cache coverage probability requirement.The simulations confirm the accuracy of theoretical derivation of the cache coverage probability.The simulation results show that the more burstiness in the traffic load,the larger the optimal BS density.When the content popularity distribution is even,increasing the cache capacity can reduce the optimal BS density.In the limited backhaul scenario,a connection-centric delivery strategy based on the on-demand allocation of the backhaul capacity is proposed.The expressions of the SCDP and energy efficiency based on the most popular caching placement strategy and the proposed content delivery strategy are derived.The simulations confirm the accuracy of the theoretical derivation of the SCDP and the energy efficiency.The simulation results show that the proposed content delivery strategy outperforms the connection-centric delivery strategy based on the average allocation of the backhaul capacity.Increasing the backhaul capacity increases the SCDP and the energy efficiency.Then,the network performance of the Hetnets caching is studied with the stochastic geometry.A small base station(SBS)deployment algorithm is proposed to meet the requirement of the cache coverage probability given the impact of the traffic load distribution on the network performance.To cope with the effect of the macro base station(MBS)having the limited backhaul capacity and the helpers lacking the backhaul links on the content delivery,a content-centric delivery strategy considering backhaul capacity allocation is proposed,and the expressions of the SCDP and the energy efficiency are derived.In the unlimited backhaul scenario,the expression of the cache coverage probability based on the most popular caching placement strategy and the connection-centric delivery strategy is derived considering the characteristics of the traffic load distribution.Given the MBS density,the optimal SBS density is obtained to satisfy the cache coverage probability requirement.The simulations confirm the accuracy of theoretical derivation of the cache coverage probability.The simulation results show that the more burstiness in the traffic load,the larger the optimal SBS density.In the limited backhaul scenario,the cooperative caching placement strategy and the content-centric delivery strategy based on the on-demand allocation of the backhaul capacity is proposed.The expressions of the SCDP and energy efficiency based on the proposed caching placement strategy and the content delivery strategy are derived.The simulations confirm the accuracy of theoretical derivation of the SCDP and the energy efficiency.The simulation results show that the proposed caching placement strategy outperforms the most popular caching placement strategy.The performance gain is obvious when the content popularity is less skewed,the cache capacity is sufficient and the helper density is modest.Finally,the network performance of the UAV assisted cellular network caching is studied with the stochastic geometry.In view of the impact of the cache serving radius on the network performance,a content-centric delivery strategy based on the cache serving radius is proposed.In addition,the expressions of the cache coverage probability,the offload probability,and the average rate are derived.In the hotspot coverage scenario,the expressions of the cache coverage probability and the average rate based on the probabilistic caching placement strategy and the proposed content delivery strategy are derived.The caching probability is optimized to maximize the cache coverage probability and the average rate respectively.The simulations confirm the accuracy of theoretical derivation of the cache coverage probability and the average rate.The simulation results show that the probabilistic caching placement strategy outperforms the most popular caching placement strategy and the random caching placement strategy.Increasing the density of the UAV deployed at higher altitude may decrease the network performance.UAV with higher density has the smaller optimal cache serving radius.In the disaster relief scenario,the expressions of the offload probability and the average rate based on the probabilistic caching placement strategy and the proposed delivery strategy are derived for the UAVs tier.The caching probability is optimized to maximize the offload probability and the average rate respectively.The simulations confirm the accuracy of theoretical derivation of the offload probability and the average rate.The simulation results show that increasing the numbers and the cache serving radius of the UAV deployed at lower altitude can improve the the offload probability and the average rate.