On The Capacity Performance Of Multi-antenna Dense Wireless Networks:A Stochastic Geomery Approach

With the development of the Internet of Things and the mobile Internet,the demand for wireless data traffic has increased rapidly.In order to realize the vision of ultra-high capacity and high data rate requirement of the next generation wireless communication system,increasing the number of access points to deploy dense networks is a promising way to enhance the reuse efficiency of spectrum resources.On the other hand,deploying multiple antennas at the access point is another way to fulfill network densification.By fully exploiting the spatial degree of freedoms to reap diversity and multiplexing gain,the spectrum efficiency of the network can be further improved.Therefore,a multi-antenna dense network is a promising technology for improving network capacity.Due to the new characteristics of multi-antenna dense networks,traditonnal anlyisis approach for network capacity may not be derictly applicable.On the one hand,with the rapid increase in the number and types of access points,the locations of access points tends to be more random.The traditional hexagonal network model cannot characterize the randonness of the topology for for dense networks.On the other hand,high channel stata information at the transmitter(CSIT)has become increasingly difficult to satisfy due increasing number of antennas and access points in dense wireless networks.CSIT inaccuracy results in inter-user interference problem that is the primary bottleneck of multi-antenna wireless networks.Taking the affomentioned features presented by multi-antenna dense networks into consideration,this dissertation aims at providing an analytical framework of multi-antenna dense networks with the help of stochastic geomery,in order to evaluate the network capacity of defferent scenarios.The goal of this dissertation is to provede insightful system design guidelines for deployment,optimization and evolution of network dendificaiton.The main contributions of this dissertation are summarized as follows:1)Considering the positions of base stations(BSs)in a dense cellular network tend to exhibit repultion,this work aims at investigating the analysis approach for capacity in a more practical network topology model,which facilitates further studying on the functional relationship between network capacity between system papameters.This work investigates the area spectral efficiency(ASE)of a two-tier network with multi-antenna BSs,whose deployment is modeled by Ginibre point process(GPP).First,using the tool of stochastic geomery,the exact and more tractable expressions of the user's average data rate are derived.With the more tractable approximation,some interesting results are observed.Analtical results demonstrate that the ASE of the networks modeled by GPP benefits more by increasing the density of small-cell BSs than that of Poission Point Process(PPP)modeled networks.In terms of the optimal number of active users for maximizing ASE,there exists a slight difference between the networks modeled by GPP and PPP.2)As multi-antenna techonogly lies in a huge demand for accurate CSIT,this has become increasingly difficult to satisfy due to the increasing number of antennas and access points in dense networks.Consiering CSIT inaccuracy results in a inter-user interference problem that is the primary bottleneck of multi-anteena networks,this work aims at investigating capacity performance for multi-antenna dense networks in the presence of CSIT errors.This work utilizes Gauss-Markov auto-regressive model to characterize CSIT errors.By employing moment-matching techniques to approximate the equivalent channel power gains,an orginal expression for ASE is derived with the tool of stochastic geometry.Based on the analytical expression for ASE,the optimization problem for ASE maximization with respect to the number of active users is considered.Analytical results demonstrate that ASE scales linearly with the number of antennas when the number of antennas is larger than a certain value.The CSIT errors determine the increasing slope.Moreover,considering inter-cell interference is a dominant factor restricting the performance of dense multi-antenna networks,this work further explores the capacity performance of coordinated beamforming in the presence of CSI delay.The user-centric interference nulling scheme is proposed.Using the tools of stochastic geomery,the approximate expression for the average data rate of users is derved.The approximate expression can be used to determine the optimal coordinated range.The impact of CSI delay on the rate performance is further investigated.3)As the next generation wireless networks tend to present a tight integration of multiple communication technologies and standards,this dissertation further considers exploit the power-domain resources.By introducing power-domain non-orthognonal multiple access(PD-NOMA)into a multi-antenna dense network,this work aims at investigating the coexistence of multi-antenna dense networks and PD-NOMA.This work developes an analytical framework for exploring the benefits of applying cluster-based multi-antenna NOMA in dense wireless networks.In the framework,the locations of BSs are modeled as a PPP for evaluating the amount of inter-cell interference.The framework relies on grouping users into different clusters,where beamforming is used to eliminate inter-group interference as well as SIC is conducted to mitigate inra-group interference.The user group strategy is formed under the framework of limited feedback.Using the tool of stochastic geometry,a new explicit expression and a tight approximation for per-group average data rates are derived in terms of relevant system parameters.Based on the analytical expressions,the impact of both spatial-domain and power-domain related system parameters(i.e.the power allocation coefficient,the BS density,the number of antennas and groups and CSIT accuracy)on throughput performance.An optimization problem to maximize ASE with respect to the number of groups and power allocation coefficient is further considered under the fariness of users.Analytical results demonstrate the performance gain of NOMA relative to orthogonal multiple access becomes marginal when the requirment of user fairness is stringent.