【作者】 刘祖浩；

【导师】 喻莉；

【作者基本信息】 华中科技大学 ， 通信与信息系统， 2013， 博士

【摘要】 无线自组织网络（Ad Hoc网络）是一类特殊的无线网络，因为其无中心、自组织、动态拓扑、多跳路由等特点，被广泛的应用于军事通信和抗险救灾等领域。另外，支持Ad Hoc通信模式也已成为无线通信设备发展的趋势。而学术界关于Ad Hoc网络的理论研究已成为了无线通信网络领域的一大分支，其中作为信息理论子课题的Ad Hoc网络的容量问题，在近十多年备受国内外广大学者的关注。本文以深入分析Ad Hoc网络容量问题为出发点，以完善网络信息理论为目标，重点研究更贴近实际网络状况的非均匀无线网络，分析了非均匀无线自组织网络的节点分布特征、网络连通性、传输模型、容量性质、传输能耗以及延时等性能。本文首先针对非均匀Ad Hoc网络的节点分布特性，引入了离散噪声过程对节点分布进行建模，其中重点分析了非均匀特性非常明显的簇扩张型网络模型，对网络中不同区域之间的节点密度差异进行了深入研究。接着本文以节点簇为基本单位，首次建立了非均匀节点分布下的渗透模型，证明了在簇内存在两种方向上的信息高速公路，由这些高速公路组成的主干传输系统成为了分析网络传输模型的关键，而在簇间则建立了相关的信息管道，为簇内的传输模型做补充，完善了整个网络的传输模型。本文还分析了该模型下，非均匀网络保证连通的充要条件，推导出了节点发射功率应满足的条件。在完善传输模型的基础上，本文提出了合理的路由传输策略，路由分为簇内传输和簇间传输两部分，文中对两部分分别进行了分析并推导证明了网络中节点传输数据的可行吞吐率，用于逼近非均匀Ad Hoc网络容量的下界。通过比较，本文所得下界较前人的结论更为紧凑。本文还重点分析了非均匀Ad Hoc网络的最小传输能量问题，旨在从理论上推导网络中随机的一对源-目的节点对之间，成功传输单位信息量数据所消耗的网络最小能量。文中通过上下界同时逼近的方法，在求取下界的过程中建立了泰森多边形模型，并在其中寻找一些较大概率不含有任何节点的空路径，通过求解一次传输所必须消耗的能量从而得到最小传输能量的下界。而在上界的求解过程中，则利用本文提出的非均匀渗透模型，计算在路由过程中所消耗的能量。通过仔细分析和证明，非均匀网络的特殊性将导致渗透模型无法完全覆盖整个网络区域，其受到网络节点分布属性的限制，最终将导致上下界存在一定的间隙。本文分析和推导了间隙的大小以及形成原因，并由此给出了非均匀网络向均匀网络转化的可行性分析及条件。最后，对于非均匀无线网络中端到端传输延时边界的问题，本文采用了随机网络演算作为理论分析工具，随机网络演算作为建模无线衰落信道，分析网络延时和数据积压的数学工具，能很好地分析网络延时的边界。本文以渗透模型中的信息高速公路为主要分析对象，通过利用网络演算中的串联定理，计算并分析了传输延时的标度率性质。结果表明，网络规模的变化对延时的影响强于对容量的影响，所以在设计大规模无线自组网的时候，可能网络的传输时延更应该引起设计者的关注。更多还原

【Abstract】 Wireless Ad Hoc network is special for its no-centre, self-organization, multihoprouting and dynamic topology, which make it widely used in areas such as militarycommunication and disaster assistance. And currently most wireless devices that appliedin our daily life can support the Ad Hoc mode. Theoretical research about Ad Hoc networkhas become an important branch of wireless communication in academic field. As asub-issue of the information theory, the capacity of Ad Hoc network has attracted muchattention in the recent decade. This thesis explores the capacity of Ad Hoc network basedon some practices of the real network. The research analyzes the performances such asnode distribution, network connectivity, transmission model, properties of capacity,transmission consumption and delay especially in inhomogeneous network, which is moreclose to the real environment.This thesis introduces the Shot Noise Cox Process to model the node distribution andanalyzes the cluster-sparse network model which is highly inhomogeneous and discussesdifferences of node density in various areas in particular. Then the thesis firstly presents apercolation model under the inhomogeneous node distribution based on cluster of nodesand demonstrates that there exist system highways in two directions. The backboneformed by the system highways is the key to analyze the network transmission model.Then the relevant “information pipes” is established between clusters to complete theintra-cluster transmission model up to the transmission model of the whole network. Thefollowing content sets the transmission power and puts forward correspondingtransmission routing scheme in line with the analysis of the sufficient and necessarycondition for connectivity. To approach the lower bound of the capacity in inhomogeneousAd Hoc network, this thesis analyzes and proves the throughput that nodes can achieve.By comparison, the lower bound obtained in this thesis is tighter than the previousresearch.Besides, to obtain the minimum energy consumption for a random pair of sourcenodes to transmit unit message, the thesis studies the minimum transmission energy of thenetwork especially. By approaching both the upper and lower bound, a tessellation model is set up when getting the lower bound to seek empty path that has no node w.h.p. to be thetransmission gap. Let the energy consumption during one transmission be the lower boundand calculate the energy consumption in the routing process using the inhomogeneouspercolation model when obtaining the upper bound. The thesis finds that there exists a gapbetween the upper and lower bound through analysis and proof. This is due to that thepercolation model can’t cover the whole inhomogeneous network under the constraint ofnode distribution.Finally, to solve the bound of transmission delay, this thesis takes stochastic networkcalculus as a tool for theoretical research and the system highway as the main analysisobject. It calculates and studies the scaling law of transmission delay using theconcatenation property of network calculus. The results show that the end-to-endtransmission delay increases nonlinearly with the increase of network size and number ofnodes. This relationship can guide the tradeoff between throughput and delay.This thesis focuses on the capacity and analyzes the performance such as energyconsumption and transmission delay. The research method is based on the informationtheory and probability theory. It first demonstrates the theorems and then does someanalysis to relevant conclusion. In view of the theoretical results, specific results havebeen given and practical meaning has been analyzed aiming at giving a guidance topractice.更多还原