Research On Resource Optimization Techniques In Edge Cache Enabled Dense Wireless Networks

To meet the growing demand of mobile data traffic in the future,both the academic and the industrial communities have carried out extensive researches on the wireless communication technologies.Among them,by deploying a large number of base sta-tions in the hot spot area,Dense Wireless Networks(DWN)has become an efficient approach to improve the capacity of wireless network in the future.However,with the large number of base stations deployed,the limited backhaul link bandwidth has become a bottleneck that limits the network capacity increase.Therefore,academic community has begun to further improve network performance by considering the characteristics of mobile user services.Popular files(such as video files)are often repeatedly requested.Such repeated transmission of files causes network congestion and increases file de-livery delay.Based on the above factors,caching popular files at base stations is con-sidered to shorten the distance between popular files and users.In this way,the file delivery delay is greatly reduced and the backhaul capacity problem is greatly allevi-ated.At present,edge cache enabled DWN has become an important network evolution architecture of the future fifth Generation(5G)technology.In edge cache enabled DWN,the existing researches mainly focus on how to im-prove the cache hit ratio.As the cache hit ratio increases,it becomes easier for the users to obtain the files directly from the nearby base stations,thus greatly reducing the file delivery delay.Besides,energy efficiency has always been an important issue in DWN.The introduction of caching can have a significant impact on the energy efficiency of DWN.In edge cache enabled DWN,file delivery delay and energy efficiency need to be balanced.In fact,both channel state and caching state can affect the user association.In other words,when edge caching is involved,energy efficiency may not be optimal in the optimal file delivery delay situation.On the other hand,a large number of caches in the base stations in DWN will also cause a certain amount of energy consumption,which cannot be ignored.To obtain the optimal file delivery delay and energy effi-ciency,we need to jointly consider file placement,user association,and power control strategies.Under the complex inter-cell interference scenario in DWN,the delay and energy consumption optimization problem will become very challenging.In addition,due to the high cost of deploying a large number of wired backhaul links in DWN,using broad-bandwidth millimeter wave as backhaul technology has been an effective means.For the millimeter Wave(mmWave)-based access and backhaul integration heteroge-neous cellular networks(mABHetNets),the introduction of edge caching will have a certain effect on spectral efficiency.Specifically,if the cache is deployed to the base station,the backhaul traffic is greatly reduced and part of the backhaul spectrum can be transferred to the access link,thus improving system throughput.In such case,it is worthy to analyze how to determine the optimal spectrum allocation between access and backhaul.For the above problems,this paper aims to optimize energy efficiency and spectral efficiency by using various resource optimization techniques.The main work and contributions of this paper are as follows:1)In the scenario where base stations cache files in DWN,this dissertation inves-tigates the tradeoff between file delivery delay and transmission energy consumption.A joint scheme of user association and power control based on a local file placement strategy is proposed to optimize the delay and energy consumption simultaneously.The analysis and simulation results show that the scheme can get the near optimal delay and energy consumption.We first analyze the file delivery delay and base station transmission energy con-sumption,respectively.And then,we establish a multi-objective optimization problem,which includes file placement,user association and cooperative power control.In prac-tice,the files cached by each base station are closely related to the preferences of users in the central area covered by the base station.Therefore,we propose a local file place-ment strategy based on the local popularity.The local file placement strategy based on local popularity allows each base station to obtain a higher file cache hit ratio.On this basis,according to the Benders decomposition method,we decompose the delay and en-ergy problem into two smaller sub-problems.We design an efficient iterative algorithm,which can achieve optimal power control and user association results by continuously narrowing the upper and lower bounds of the objective value.2)Considering the impact of the cache strategy on delay and power consumption,this dissertation establishes the joint optimization model of systemic total file delivery delay and systemic total power under a global file placement strategy.A user association and power control algorithm based on the global file placement strategy is proposed.Simulation results show that the algorithm can achieve near-optimal performance of total system delay and total power consumption.Considering the caching power consumption,this dissertation models the file de-livery delay and power consumption optimization problem as a mixed integer program-ming problem.According to the generalized Benders decomposition method,we can decompose the original problem into the primal problem related to the transmission power allocation and the master problem related to the file delivery path selection.The transmission power allocation problem is solved by power control and the file delivery path selection problem is solved by file placement and user association.Then,based on generalized Benders decomposition,we propose an iterative algorithm.Finally,we can get the near-optimal file placement,user association and power control results.3)In Dense Wireless Heterogeneous Networks where access link and backhaul link share millimeter wave spectrum,this dissertation considers the impact of caching on spectrum efficiency.When edge caching is involved,this dissertation analyzes how to determine the optimal spectrum allocation between access and backhaul to improve system throughput and spectral efficiency.Simulation results verify the correctness of theoretical analysis.For mABHetNets,if the cache is deployed to the small base stations,the backhaul traffic used for delivering files is greatly reduced.At this time,the corresponding back-haul spectrum can be transferred to the access link,thus improving system throughput.Therefore,when cache is involved,we need to analyze how to determine the optimal spectrum allocation between access and backhaul.We first obtain a network analysis framework by using stochastic geometry.Based on this theoretical framework,we ob-tain the channel models of line-of-sight and non-line-of-sight propagation of millimeter wave.According to the definitions of average potential throughput and area spectral ef-ficiency,we derive the corresponding expressions.Then we analyze the effect of cache capacity allocation and spectrum bandwidth allocation on network performance.Ac-cording to the analysis results,we find that different popularity parameters and caching power consumption coefficients will affect the area spectral efficiency.At this time,considering the high complexity of the area spectral efficiency expression,we further derive the simplified approximate expressions in noise-limited and interference-limited cases,respectively.Simulation results show that the caching can improve the average potential throughput and area spectral efficiency of the network.In addition,through the simulation result,we know that with a higher popularity parameter and a lower caching power consumption coefficient,a higher area spectral efficiency can be achieved.