Research On User Association Methods In Heterogeneous Networks Modeled By Stochastic Geometry

Recently,terminal equipments such as smart phones,wearable devices and tables are rapidly gaining popularity in our daily lives,and massive quantities of these terminal equipments require to aceess the networks.Meanwhile,all kinds of multimedia services such as online game,mobile social network service,virtual reality and augmented reality have also sprung up,so the demands of quality of service(Qo S)for users are constantly improving.However,the traditional single-tier cellular networks have been unable to cope with the network access demands of massive terminal equipments and the exponential explosion of mobile communication traffic.Therefore,the network architecture is evolving toward the heterogeneous networks(Het Nets)that are randomly deployed,diversified and overlapping covered by different types of cells.In the Het Nets,user association methods are the keys to meet the demands of Qo S for users and improve the overall network coverage,spectral efficiency and network capacity,especially several new technologies such as non-orthogonal multiple access(NOMA),millimeter wave(mm Wave)communications and ultra-dense networks(UDNs)will be integrated into Het Nets.Hence,the traditional user association methods need to be investigated further.In this paper,the user association methods in Het Nets will be the key issue in order to improve the overall network spectral efficiency and guarantee the Qo S of users.The randomness and heterogeneity of Het Nets will be modeled by stochastic geometry,which is a new mathematical statistical tool for spatial point distribution.The content of this paper is to design the user access methods for Het Nets from four aspects,including load balancing,interference coordination,Het Nets using hybrid multi-access technology and mm Wave communications,and ultra-dense networks.The general mathematical expressions of coverage probability,average user rate and network throughput are deduced theoretically,and the effects of the designed user association methods and the proposed algorithms on network performance are verified through theoretical analysis and simulations.The specific research contents are as follows:Firstly,for the load balancing of two-layer Het Nets,the cell range expansion(CRE)technology is used to artificially offload the users from the crowded macro to the pico.For the offloded users and pico edge users are vulnerable to suffer from the excessive cross-tier interference from adjacent macros,a soft frequency reuse(SFR)algorithm is adopted.In detail,the whole spectrum are divided into two bands and the power control is applied to the macro interior users,such that improve the network performance and guarantee the Qo S of pico edge users.Based on the stochastic geometry method and considering the random resource allocation and the practical cell load model,the analytical expressions of network coverage probability and average user rate are derived,which can be reduced to closed-form in some case.At the same time,the theoretical and simulation results are provided to analyze the impact of the bias,resource partition coefficient,power control coefficient and SFR threshold on the network performance.Compared with the traditional resource partitioning algorithm,the simulation results show that the CRE and SFR can effectively improve the average user rate.Secondly,for the limited offload capacity of the pico nearby macro,the edge deployment of small cell is adopted,that is,the pico is artificially deployed in the edge area of macro to guarantee the Qo S of macro edge users.For the load balancing and interference coordination in Het Nets,the CRE and coordinated multipoint transmission are adopted.The cooperation threshold can be regarded as the bias of CRE and the judgment criteria for cooperative transmission of offloaded users.At the same time,based on the stochastic geometry method,the analytical expressions of coverage probability and average user rate of a typical user are derived.The theoretical and simulation results are provided to analyze the impact of the radius of macro inner region and cooperation threshold on the network performance.The simulation results prove that the proposed algorithm can guarantee the Qo S of macro edge users and improve the overall network performance.Thirdly,in order to further improve the spectral efficiency of Het Nets,NOMA is adopted in the small cells with considering distance-based ordering and instantaneous signal-to-interference-plus-noise ratio(ISINR)-based ordering schemes.To meet the high data rate requirements of hot users,the femto BSs with mm Wave communications are deployed.In the hybrid Het Nets using hybrid multi-access technology and mm Wave communications,the CRE is used to balance the load among each tier.At the same time,based on the stochastic geometry method,the analytical expressions of coverage probability and average user rate of a typical user are derived.The theoretical and simulation results are provided to analyze the impact of the bias,two different user ordering schemes,and NOMA power allocation fraction on the network performance.The simulation results prove that the the hybrid Het Nets with CRE can meet the demands of high Qo S and data rate of hot users.Finally,one of the future evolution trends of Het Nets is the ultra-dense networks(UDNs).Considering the line-of-sight(LOS)and non-line-of-sight(NLOS)propagations,the NOMA with distance-based ordering and received instantaneous signal power(RISP)-based ordering schemes are adopted to improve the whole network spectral efficiency.The closest BS association rule with low complexity is also considered.At the same time,based on the stochastic geometry method,the analytical expressions of accuracy probability,coverage probability and network throughput are derived.The theoretical and simulation results are provided to analyze the impact of BS density on the accuracy probability,and the impact of BS density,two different user ordering schemes,and NOMA power allocation fraction on the coverage probability and network throughput.The simulation results prove that the UDNs with NOMA can improve the network throughput.