基于随机几何的异构蜂窝网络性能分析

发布时间：2018-01-12 22:37

本文关键词：基于随机几何的异构蜂窝网络性能分析　出处：《中国科学技术大学》2016年博士论文　论文类型：学位论文

【摘要】：爆炸性移动数据业务的增长使得传统同构蜂窝网络结构难以应对数据量的冲击以及用户对于通信质量的要求,而异构蜂窝网络架构则广泛被认为是应对该挑战的有效方式之一。蜂窝网络密集化、多元化和随机化的异构趋势导致网络架构愈发复杂,从而使得传统的网络性能评估方法因其过于复杂或相关假设过于理想而无法适用于异构蜂窝网络。然而,对网络进行相应的性能评估是决策网络参数设定和引入新型网络部署和技术的前提条件。因此,找寻一个有效且便于分析的网络性能评估方法是当前异构蜂窝网络研究领域中亟待解决的问题之一。目前随机几何理论是广泛用于建模和分析网络的重要数学工具,且所得理论结果与实际仿真结果切合。因此,本文考虑不同类型点过程建模的异构蜂窝网络场景,利用随机几何相关理论分析不同的关键网络性能指标,期望给予可分析性较高的理论结果,从而在理论上了解网络技术和参数对于网络性能指标的影响。首先,考虑最简单的泊松点过程所建模的异构蜂窝网络,分析其最基本的性能指标信干比分布和可达速率以及更全面反映网络性能的指标,如区域频谱效率和网络能效。由于模型的可分析性较高,本研究考虑更为实际的多信道环境下的中继异构蜂窝网络,采用泊松多层异构蜂窝网络模型对基站和中继的空间分布进行建模,并给出网络性能的理论分析,进而研究站点密度等网络参数对于网络性能的影响。在该模型下,本文推导了宏用户和中继用户的覆盖率及相应的平均可达速率,并进一步推导出区域频谱效率和网络能效的理论闭式表达式。利用齐次泊松点过程的独立稀释特性,本文通过以一定概率去关闭或者开启站点以建模站点密度的动态特性,从而探究网络性能与站点密度之间的关系。结果显示中继异构蜂窝网络的性能要明显高于传统单层蜂窝网络,其中随着站点密度的增加,区域频谱效率随之增加而网络能效先增加后减小,因而站点密度对于区域频谱效率和能效的影响存在一定折中。简单网络模型所分析的结果与方法为更复杂一般性的模型分析提供了基本的理论结果和参考方法。其次,将建模异构蜂窝网络的空间点过程推广到一般性的非泊松点过程,提出有效且简单的分析方法,进一步推广更为实际的非泊松模型在网络研究上的应用,从而得出更具有实际意义的理论结果。以最基本的性能指标信干比分布为研究目标,本文提出两种适用于非泊松多层网络的泊松近似方法,以给出信干比分布较为精确却简单易分析的理论结果,如此可兼顾模型的精确性和可分析性。本文首先对一般性异构蜂窝网络建立了每层泊松近似方法,并基于每层泊松近似方法给出适用于所有路损指数相等的场景中的有效增益泊松近似方法,即推导出异构蜂窝网络的有效增益的具体表达式,如此相应的信干比分布则可以通过以泊松网络简单的信干比分布为基础,在信干比门限上平移有效增益即可。此外,两种泊松近似方法也可用来刻划一般性异构蜂窝网络的信干比分布的拖尾特性。更进一步地,本文推导了在β-Ginibre和泊松点过程所组成的双层异构蜂窝网络场景下精确的信干比分布,通过在精确度和表达式简易性两个方面比较了精确理论结果和近似结果以说明所提近似方法的有效性。结果证明所提的两种方法可以给出简单且精确的信干比近似分布,因此本研究为进一步分析其他更能反映网络性能提供了基础分析结果。最后,在非泊松多层网络模型下,利用近似信干比分布,进一步考虑其他更能反映网络性能的指标以及相关技术对网络性能的影响。同样考虑区域频谱效率和网络能效两种性能指标,并基于由泊松近似方法所得来的近似信干比分布研究一般性异构蜂窝网络场景下的频谱分配策略对于这两种性能的影响,以确定不同频谱分配策略在网络中的应用范围。分别以最大化这两种性能指标为目标,优化各自在一定频谱策略下各层所需频谱大小和各层信干比门限。首先考虑频谱分隔策略,并形成相应的目标性能优化问题,在此基础上引入不同的频谱效率限制条件。该优化问题可等价分解为两个子问题,其中一个是获取最优的信干比门限值以最大化相应的频谱效率,另一个是基于最优信干比门限给出最优的层间频谱分配策略以最大化区域频谱效率或网络能效。随后,利用异构蜂窝网络的有效增益泊松近似方法,本文证明了频谱共享优化问题可以转化为频谱分隔优化问题,因此所提的解决方案亦适用于频谱共享问题。此外,本文比较了泊松模型和非泊松模型在采用不同频谱分配策略下的网络性能,且发现在两种模型下这两种性能随着网络参数的变化呈现类似的变化趋势。本文研究结果为研究更为复杂多层异构蜂窝网络以及评估新技术新方法对于网络性能的影响提供了一种可行性较高的理论方法,具有重要的理论指导意义和实际应用价值。
[Abstract]:The explosive growth of mobile data services makes the traditional cellular network structure is difficult to cope with the isomorphism of the amount of data and the impact of user requirements for the quality of communication, and heterogeneous cellular network architecture is widely considered to be one of the effective ways to cope with the challenges. The cellular network intensive, heterogeneous diversification and randomization leads to more complex network architecture, which makes the the traditional method of network performance evaluation because it is too complex or too ideal assumptions are not suitable for heterogeneous cellular networks. However, to evaluate the performance of the corresponding network is a prerequisite for making network parameter setting and the introduction of a new network deployment and technology. Therefore, looking for an effective and easy to analyze network performance evaluation method is one of the urgent heterogeneous cellular network in the research field. At present the random geometry theory is widely used in modeling The network and analysis of important mathematical tools, and the simulation results and the theoretical results fit. Therefore, we consider heterogeneous cellular network scenarios for different types of process modeling, the key performance indexes of different network analysis using stochastic geometry theory, theoretical analysis results can be expected to give high, so as to theoretically understand network technology and effect of parameters on the network performance. Firstly, considering the heterogeneous cellular network the most simple Poisson process modeling, analysis of the basic performance index of SINR distribution and achievable rate and more fully reflect the network performance indicators, such as regional spectrum efficiency and network efficiency. Due to the analysis of the model is high. This study considers the relay heterogeneous cellular network multi channel environment is more practical under the multi heterogeneous cellular networks using spatial Poisson model to the base station and the relay. The distribution model, and analyzed the network performance theory, then studies the network parameters for site density affect the performance of the network. In this model, this paper deduces the coverage of macro users and relay users and the corresponding average rate, and further deduce the area spectral efficiency and energy efficiency of the network theory of closed expression the use of independent dilution characteristics. The homogeneous Poisson point process, this paper with a certain probability to close or open the site to build the dynamic model of site density, so as to explore the relationship between network performance and site density. The results show that the performance of the relay heterogeneous cellular network are significantly higher than the traditional single cellular network, which increased with the site density. Area spectral efficiency is increased and the network efficiency increased first and then decreased, thus the site density influences on regional spectral efficiency and energy efficiency of the existence of a Make compromise. Provide theoretical results and reference method for basic results of simple network model and analysis method for more complex general model. Secondly, the spatial modeling for heterogeneous cellular networks is extended to non Poisson point process of general process, put forward analysis method is effective and simple, to further promote more application non Poisson model in the actual network research, thus obtains the theoretical result has more practical significance. With the performance index of the basic stem ratio distribution channel as the research object, this paper puts forward two kinds of Poisson for non Poisson multilayer neural network approximation method to give the signal to interference ratio distribution theory results more accurate but simple and easy to analyze so, can take into account the model accuracy and analysis. Firstly, the general heterogeneous cellular network established for each layer Poisson approximation method based on Poisson approximation method for each layer Suitable for effective gain all the path loss index equal to the Poisson approximation method in the scene, the specific expression of effective gain is presented for heterogeneous cellular networks, so the corresponding SNR distribution by Poisson network simple SINR distribution based on stem gate limit in the letter translation can be effective gain. In addition, two kinds of Poisson approximation methods are also used to characterize the general nature of heterogeneous cellular network SNR smear characteristic distribution. Further, this paper presents the precise double channel heterogeneous cellular network scene composed of beta -Ginibre and Poisson point process under dry ratio distribution, the effectiveness of the proposed method in the approximation two accuracy and simple expression comparison of precise theoretical results and approximate results to illustrate. Results show that the two proposed methods can give a simple and accurate approximation for SNR distribution. This research for the further analysis of other more can reflect the network performance provides the basis of the analysis results. Finally, the non Poisson multilayer network model, using approximate SINR distribution, further consider other more can reflect the influence of network performance indicators and related technology on the performance of the network. Also consider the area spectral efficiency and network energy efficiency two the performance index, and based on the approximate Poisson approximation method to the letter by the effect of general heterogeneous cellular network scenarios than the distribution of spectrum allocation strategy for the two kinds of performance, to determine the scope of application of different spectrum allocation strategy in the network respectively. In order to maximize the two performance targets and their optimization in certain frequency strategy of each layer and each layer to the size spectrum SNR threshold. We first consider the spectrum separation strategy, and the formation of the target performance of the corresponding optimization problem, in On the basis of introducing limited spectrum efficiency of different conditions. The optimization problem can be decomposed into two sub problems, one of which is to obtain the optimal SNR threshold to maximize the spectral efficiency accordingly, another is the optimal SNR threshold gives the optimal spectrum allocation strategy between layers to maximize regional spectrum the efficiency or the network energy efficiency. Based on then, effective use of heterogeneous cellular networks gain Poisson approximation method, we prove that the spectrum sharing optimization problem can be transformed into a separate spectrum optimization problem, so the solution proposed is also applicable to the spectrum sharing problem. In addition, this paper compares the Poisson model and non Poisson model in network performance with different spectrum the distribution strategy, and found that under the two models of the two performance with the change of network parameters showed similar trend. The results of this study are to study more complex A multi-layer heterogeneous cellular network and evaluation of new technologies and new methods for network performance provide a highly feasible theoretical method, which has important theoretical significance and practical application value.

【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN929.53

【相似文献】