Investigations On Deployment Optmization Of Ultra-dense Wireless Networks

It is predicted that there will be 1000 times in the mobile data traffic over the next decade.Improving the performance of the wireless communication network system has become the focus of academia and industry.Ultra-dense network is a promising way to meet the Requirement.Optimal deployment of ultra-dense network is a key problem.In this thesis,the optimal deployment of ultra-dense network is invistigated.The deployment optimization of two non-unifrom ultra-dense networks is investigated.Different from FULL BUFFER traffic adopted in existing literatures,non FULL BUFFER traffic is adopted.Closed form expressions of SINR and Rate performance of the two models are derived,and simulation results are presented to verify our derivation.For a two-tier ultra-dense heterogeneous network,different from existing literatures,the network performance with non FULL BUFFER is investigated.In this network,macro base stations are located following homogeneous Poisson Point Process(PPP),pico base stations are located following Poisson Hole Process(PHP),and users are located according to PPP.This deployment lead to reduced number of pico base stations near the macro base stations.In addition,in this scenario,for a given number of pico base stations,the density of pico base station with larger distance from the macro base station is larger,thus can improve the data rate of the pico user.User's data traffic arrives following Poisson process,and each user chooses serving base station based on the maximum RSRP criterion.Users of the same base station are queued to wait for the service according to the ararival time.Queuing theory and stochastic geometry theory are used to derive the close-form expressions of SINR and Rate distribution with non FULL BUFFER.Monte Carlo simulations are conducted to verify the close-form expressions.Numeric simulation of the performance indicators,such as rate,power consumption and transmission probability,for FULL BUFFER and non FULL BUFFER traffic are presented.Results revealed the advantages of the model.An improvement to the network model is proposed.In the proposed model,macro base stations are located according to PPP,pico base stations are located according to Poisson Cluster Process(PCP).The packet of each user arrives following Poisson prcess.Non FULL BUFFER traffic is considered.Users located in hot spot regions connect to pico base station,and others connect to macro base station.PCP deployment more applicable for office buildings,shopping malls and other hot spot regions where a large number of users gather and have great demand of data traffic.Deployment of more pico base stations with higher density in hot spot regions,and macro base staion in the rest area to provide coverage can improve the user experience and system throughput.The expressions of SINR and Rate distribution are derived based on queuing theory and stochastic geometry,and Monte Carlo simulation is presented to verify these close-form expressions.