Research On Performance Analysis And Energy Efficiency Optimization For Ultra Dense Networks

The fifth generation mobile communication system(5G)is expected to play a key role in satisfying the challenges of explosive growth of wireless data traffic,massive connections of smart devices,continuous emergence of new services and applications in the future.Particularly,in the hierarchical heterogeneous architechture,ultra-dense netwok(UDN),which significantly enhances the system capacity by deploying ultra-high density access points,is considered as one of the key technologies for 5G system.UDN generally consists of a large number of different small base stations(BSs)with low power and low cost,whose deployment density is much higher than current mobile communication networks.As these dense access points can shorten the distance between receivers and senders and reduce the transmission path loss,the transmission quality can be improved effectively.Moreover,by refusing spectrum in all kinds of base stations,the system capacity and the spectral efficiency can also be greatly improved.However,in order to fully exploit the advantages of UDN,there are some problems still need to be addressed.These problems can be summarized as two aspects: the performance analysis before UDN deployment and the performance optimization after UDN deployment.The performance analysis is mainly to construct the spatial point model of UDN nodes,and to analyze the influence of different network configuration parameters on the system performance,so as to provide a theoretical basis for the deployment of UDN.The performance optimization mainly uses convex optimization,game theory or other tools to optimize network performance under the condition that the network configuration parameters have been determined.Bsed on these two aspects,this thesis uses stochastic geometry to model and analyze the perfomence of UDN,and then design energy efficiency optimization schemes based on resource allocation and base station sleep to improve the performance of UDN.For the performance analysis before UDN deployment,this thesis employs two layers of UDN as an example to model the distribution of pico base stations and femto base stations into a homogeneous poisson point process and a Matern cluster process respectively.By using stochastic geometry,the coverage rate and mean achievable rates of users are derived.Furhtermore,the area spectral efficiency and area energy efficiency of UDN are also deduced.Through the theoretical analysis and the system level simulation,the relationship between the system parameters and the network performance is obtained.For improving the performance of UDN,a cluster based energy efficient resource allocation scheme is proposed.The scheme consists of two stages called clustering stage and resource allocation stage.In the stage of clustering,a modified K-means clustering algorithm is designed.it can automatically adjust the number of clusters according to the base station density.Then,in each base station cluster,the user grouping algorithm is designed with the purpose of minimizing interference within each base station cluster.In the resource allocation stage,a spectrum allocation algorithm and a game based power allocation algorithm are designed.Simulation results show that the proposed scheme can effectively improve the throughput,spectrum efficiency and energy efficiency of the system,and this advantage becomes more and more obvious as the base station density increases.In addition,this thesis also proposes a two-stage UDN base station sleep scheme.By using stochastic geometric theory,the small base stations and users in the hot spot are model as two separate poisson points processes.Hence,the mean achievable rates of users and the system energy consumption of traditional and random sleep mode can be got.Compared with the exhaustive search method,the proposed scheme can greatly reduce the complexity.Moreover,compared with the traditional sleep mode and random sleep mode,the proposed scheme has obvious improvement in energy efficiency.