Research On Spectrum Management And Performance Analysis In Dense Cellular Networks

Dense deployment is one of the most important evolution trends for future mobile cellular networks.With a huge amount of small base stations added into the networks,the topology and structure of cellular networks have appeared to be quite different than in the past.The difference mainly lies on that the distribution of the base stations is changing towards an irregular manner than the traditional regular spatial distribution.Moreover,the dynamic "on" and "off" of the small base stations will lead to the dynamic changes of the network topology and structure in some local areas.Hence,there are new requirements for introducing self-organizing management techniques into the large-scale dense cellular networks and developing novel tools and methods for network modeling and performance analysis in the future.Among the various studies on the self-organization oriented network management,spectrum management is one of the most important techniques.The dynamic of the dense cellular networks in local topology makes it reasonable to perform the spectrum management in the coverage cells of each individual base stations in a distributed and self-organizing manner,so that the spectrum resources allocated to the base stations can adapt to the variations of the network topology timely and providing guaranteed ser-vice qualities to the users.Meanwhile,for the further improvement of the spectrum efficiency under the shortage of the spectrum resources due to the network densification,spectrum sharing between different mobile network operators is deemed to be a potential and promising network evolving direction.However,the design of the system structure supporting cross-operator spectrum sharing,together with the cross-operator interference coordination and performance protection for the user on the licensed band of each oper-ator,are the key issues to be solved.For the future large-scale dense cellular networks,due to the lack of sophisticated network pre-planning,the network topology and structure will lose the advantage of spatially regular distribution in traditional cellular networks.At this time,how to effectively model the distribution of the base stations and users and analyze the user performance for providing guidance to the deployment in real world are still problems demanding much more study.This dissertation focuses on the spectrum management and network modeling and performance analysis problems in dense cellular networks,and the main contributions are:1)For the self-organizing management in dense cellular networks,we focus on the ones with dynamic changes in local topology.We analyze the difficulties and re-quirements in spectrum resource allocation and algorithm design,and build up a criterion for determining the interference coupling relationship between two fem-tocells.Based on that criterion,we design a distributed and autonomous carrier selection scheme according to the "clique" concept in graph theory for the femto-cells in the local area.This carrier allocation scheme is able to adapt to the variation of the network topology.In the end,a guarantee mechanism for the quality of ser-vice based on the interaction among femtocells is developed to fulfill the minimum rate requirement from each femtocell as much as possible.2)For the self-organizing management in dense cellular networks,by focusing on the coexistence problem of different operators in the same local area,we propose a specific system architecture which supports the spectrum sharing across different operators,and discuss the functionalities of the entities.Then we combine the earlier distributed autonomous spectrum allocation scheme with an operator prior-ity guarantee mechanism and propose the cross-operator spectrum sharing scheme based on the primary-secondary priority.Such a sharing scheme can effectively guarantee the individual operators’ network performance on their licensed bands,meanwhile offering them more opportunities to get access to the secondary bands.3)For the large scale single-tier dense cellular networks in the future,we model the uplink and downlink cellular networks based on OFDMA and the point process the-ory in stochastic geometry,and discuss the Poisson point process approximation of the user distribution.Then we further perform the performance analysis of the meta distribution with fractional power control,characterizing the user performance dis-tribution via an analytical way.We derive the moments of the conditional success probability of the SIR(Signal-to-Interference Ratio)and obtained several impor-tant performance metrics such as its mean(i.e.,the well-known standard success probability or coverage probability),variance and mean local delay.The analyti-cal results reveal more fine-grained information about the effect of fraction power control on both the uplink and downlink.4)For the heterogeneous scenario where the cellular networks coexist with D2D(De-vice to Device)communications in future large-scale dense cellular networks,we conduct the performance analysis based on the meta distribution for the overlay and underlay deployment scenarios respectively,with the receivers in capable of per-forming interference cancellation.For the former,we derive the general expression for the b-th moment of the conditional success probability for the typical D2D user,together with the upper and lower bounds and the asymptotics of the first moment and the variance of the meta distribution.Then we study the horizontal shift phe-nomena between the b-th moment curves under different numbers of the canceled interferers by explicitly providing the asymptotic horizontal gain expression.Fur-ther,by utilizing the mapping between the meta distribution of the SIR and the meta distribution of the achievable rate,we study the effects of the spectrum division ra-tio on the distribution of the achievable rate reliabilities and discuss the selection of the optimal ratio.For the latter scenario,we derive the exact expressions of the b-th moment for the SIR meta distribution of the cellular and D2D users,respec-tively.Furthermore,we study the effects of the densities of the base stations and D2D links on the SIR meta distributions,and also the horizontal shift property of the b-th moment curves of their users under perfect interference cancellation.