Performance Analysis And Optimization In Large-scale Distributed Mobile Networks Under A Stochastic Geometry Model

Along with the rapid development of mobile communication industry and the gradual application of the fifth generation mobile communication system(5G),the demands of mobile users and mobile devices keep growing and large-scale distributed mobile networks,as the key network architecture in 5G system,possesses much higher macro-diversity gain and smaller transmission distance.In addition,large-scale distributed antenna system(DAS)can improve system capacity and spectral efficiency(SE)by adding multiple remote radio units(RRU).Large-scale distributed antenna system can improve the performance,but produce more backhaul power consumption and cause interference management problem.To analyse the system performance easily,we take advantage of stochastic geometry model and consider user-centric clustering against the interference,then we analyse and optimize the performance of large-scale distributed mobile networks.Firstly,we introduce the channel model of wireless communication,including path loss model,shadow fading model and small-scale fading model,then multiple input multiple output(MIMO)system model and channel capacity are briefly described.Furthermore,we introduce the poisson point process(PPP)and its properties,and analyse the performance of the system under a stochastic geometry model like PPP.Subsequently,we consider a downlink large-scale distributed antenna system in which the RRUs are connected to central processing unit via rate-limited backhaul links,then investigate the antenna assignment and power optimization under this system model.We assume a fixed signal to interference plus noise ratio(SINR)constraint for each user,and formulate a optimization problem to find the tradeoff between the sum transmit power and the sum backhaul power consumption.This thesis take advantage of the l0-norm and l₁-norm to transform the problem,and use the iterative weighted l₁-norm algorithm and uplink-downlink duality algorithm to solve the problem.Moreover,the simulation results show that uplink-downlink duality algorithm has low computational complexity and converges in a small number of iterations.Then,we investigate the downlink ergodic rate in large-scale distributed antenna system under stochastic geometry model.Users and RRUs are modeled by PPP,and we make use of distance-based user-centric clustering.Moreover,maximum ratio transmission(MRT)precoding are introduced into each virtual cell.We derive the expression of downlink ergodic rate by using the properties of poisson point distribution,and obtain the Laplace transform expressions of desired signal and interference signal.Furthermore,compared with Monte Carlo simulation results,the numerical simulation results show that the approximate expression provides a tight upper bound,and the increase of virtual cell size can improve the ergodic capacity of users.Finally,as the number of RRUs grows,the energy consumption problem become increasingly serious and energy efficiency(EE)indicators have become important indicators to measure communication systems,so we focus on the optimization of the energy efficiency in large-scale mobile networks.We introduce a new definition of coverage probability under stochastic geometry theory,and derive a new formulation of system spectral efficiency(SSE),which depends on the transmit power and deployment density of the RRUs.Then a tractable closed-form formula of system energy efficiency(SEE)is derived.It is mathematically proved,given the density of RRUs,the system energy efficiency increased firstly and then decreased in the transmit power and exists the optimal transmit power.Simulation results show that given the transmit power,a unique density of RRUs exists,which maximizes the system energy efficiency and there are optimal clustering results in large-scale distributed antenna systems.