共查询到20条相似文献,搜索用时 187 毫秒
1.
容错性是多计算机网络中非常重要的研究主题.本文基于节点随机出错概率研究多计算机网络Mesh的容错性,采用子网划分方法,将网络划分为相互独立且不相交的子网,假设每个节点具有随机出错概率,通过分析子网的连通性,得到整个网络的连通概率.数值和模拟结果表明,网络连通概率随时间的增大而减小,在给定的时间内,网络规模越大,连通概率越低.例如,对于给定的指数分布(λ=3 509×10-6),当时间比较小(4000秒内)的情况下,多达四万节点的Mesh网络几乎总是连通的,连通概率达到99%以上,这也表明以Mesh网络为拓扑的多计算机系统是相当可靠的. 相似文献
2.
3.
4.
针对车载自组织网络(VANET)的连通性问题,分析了其拓扑动态变化过程中网络的连通性的演化特征。首先,提出以连通分支数、连通概率及连通长度为评价指标的VANET拓扑连通性参数;然后,结合车辆换道功能的智能驾驶移动模型(IDM-LC),应用VanetMobiSim仿真软件建立VANET;最后,通过仿真实验分析了节点通信半径与平均连通分支数、平均连通率及平均连通长度之间的关系,同时分析了VANET连通分支数的统计分布特征,用Q-Q图和T检验验证得出结论:连通分支数服从正态分布,且该统计分布特征与节点通信半径无关。 相似文献
5.
基于齐次泊松点过程的节点分布模型,在不同的衰落信道模型下推导了网络无孤立点概率的闭型表达式,用作网络连通概率的上界。特别分析了对数正态阴影衰落和瑞利衰落信道的相关物理参数对连通性的影响。此外,还讨论了协作通信对于网络连通性能的提高作用。最后,在仿真构造的无线网络中测试得到的连通性能的仿真值与理论分析结果吻合。 相似文献
6.
7.
8.
分析连通支配集的支配性约束和连通性约束条件,提出2条针对简单无向连通图最小连通支配集问题的化简规则.规则通过对图中节点的邻节点进行分类以及寻找图的割点提前确定一些必选节点,同时删除一些多余节点,从而降低原问题的规模.从理论上证明了化简规则的正确性,并通过随机仿真实验验证化简规则的有效性. 相似文献
9.
10.
无线传感器网络的初始配置最优可以减少传感器网络的拓扑变化和降低网络重置的能量消耗.对初始均匀随机分布的无线传感器网络的连通性进行了研究.运用覆盖理论给出了传感器节点的连通度概率分布模型,并在此模型基础上推导出传感器节点的通信半径与期望连通度概率最大之间的关系.仿真结果表明了结论的正确性. 相似文献
11.
针对在低阶脑网络应用图论忽视了功能连接高阶动态性的问题,提出了一种基于高阶动态功能连接的图论网络构建方法(GNC-HodFC),提取高阶FC网络的图论特征以对轻度认知障碍患者和健康被试者进行差异性分析及分类。首先定义了表征高阶动态脑网络连接的图论节点和边;然后利用滑动窗相关技术提取低阶功能连接信息,提出平稳性判据,选取最优特征子集以构建图论的节点;最后提出自适应阈值策略对高阶动态功能连接信息进行选取以构建图论的边,最终完成高阶动态脑网络的图构建。实验结果表明,GNC-HodFC的平均分类准确率可以达到70.5%,优于其他三种对比方法,且患者组和健康组的图论特征中存在显著性差异,GNC-HodFC方法可以为轻度认知障碍的诊断提供新的辅助手段。 相似文献
12.
13.
关于单节点修复模型,Dimakis已通过信息流图分析出节点存储与修复带宽的理论界。对于多节点的修复,Shum和Hu提出了新节点之间相互合作的模型,并给出此模型下存储与带宽的理论界;Zhang等人介绍的新节点之间不再传输数据的模型,比合作修复减少了设计和运算的复杂性,更符合系统的需要。针对这种新模型,利用割型找出其最小容量割,并用线性规划的方法给出存储—带宽的理论界,过程更为简单。最后给出一些特殊参数下的编码构造。 相似文献
14.
在真实的网络环境中,很多节点可能是自私的,它们不愿意牺牲自己的资源为其他节点转发消息。针对这种情况,提出一种基于博弈论的激励机制,可以激励节点与其他节点相互合作。该机制为二阶段激励,激励节点接收消息以协助其他节点转发,同时激励节点转发更多的消息。把源节点与中继节点之间的竞争与合作模型化为Bertrand(伯特兰德)博弈,定义了源节点和中继节点的效用函数。求解了源节点的最佳定价策略和中继节点最佳的转发计划,验证了源节点与中继节点之间存在唯一的纳什均衡。模拟仿真结果表明提出的激励机制能够鼓励自私节点参与合作,能提高路由算法的传递率,同时降低了消息传递延迟。与基于声誉的激励机制相比,所提激励机制能使消息传递成功率提高31.4%、平均时延降低9.7%。 相似文献
15.
Ahmed M. KhedrAuthor Vitae Walid OsamyAuthor Vitae 《Journal of Parallel and Distributed Computing》2011,71(10):1318-1326
Target tracking is an important sensing application of wireless sensor networks. In these networks, energy, computing power, and communication bandwidth are scarce. We have considered a random heterogeneous wireless sensor network, which has several powerful nodes for data aggregation/relay and large number of energy-constrained sensor nodes that are deployed randomly to cover a given target area. In this paper, a cooperative approach to detect and monitor the path of a moving object using a minimum subset of nodes while maintaining coverage and network connectivity is proposed. It is tested extensively in a simulation environment and compared with other existing methods. The results of our experiments clearly indicate the benefits of our new approach in terms of energy consumption. 相似文献
16.
In wireless ad hoc networks cooperation among nodes cannot always be assumed since nodes with limited resources and different
owners are capable of making independent decisions. Cooperation problems in topology control and packet forwarding tasks have
been mostly studied separately but these two tasks are not independent. Considering a joint cooperation problem by taking
into account dependencies between tasks will result in more reliable and efficient networks. In this paper topology control
definition is extended to cover cooperation problem in both packet forwarding and topology control in a single problem. In
this definition nodes have to adjust their transmission power and decide on their relay role. This paper models the interactions
of nodes as a potential game with two-dimensional utility function. The presented model, named TCFORCE (Topology Control packet
FORwarding Cooperation Enforcement), preserves the network connectivity and reduces the energy consumption by providing cooperative
paths between all pairs of nodes in the network. 相似文献
17.
The number of hops between source node and destination node is a key parameter in studying multi-hop wireless networks. Although hop count in wireless ad-hoc networks (AHNs) has been studied in the literature, no works on investigating the hop count characteristics in cognitive environments have been carried out. In this paper, we model cognitive radio ad-hoc networks (CRAHNs) as geometric random graphs and then propose a framework for studying the hop count distribution and correlated connectivity of communication path between two arbitrary nodes in CRAHNs with shadow fading. The framework consists of an algorithm and a methodology. Specifically, from the perspective of geometric random graph, the algorithm finds all possible paths between two arbitrary nodes and returns the hop count of the shortest path between them by using the global location information of nodes, i.e. primary users – PUs and secondary users – SUs, and the active states of PUs as input data. Meanwhile, through huge number of random network topology trials, the methodology returns the hop count distribution and connection status of communication path between two arbitrary nodes in CRAHNs with shadow fading. From the evaluating scenarios in this paper, important features of hop count distribution and connectivity and their correlating relationship in CRAHNs with shadow fading are revealed and compared with those in AHNs and in CRAHNs without shadow fading. 相似文献
18.
We lay down the foundations of a new approach for finding the network connectivity in wireless networks, with special regard
to the properties of dependencies between links of geometrically collocated nodes. The proposed methodology is rooted in the
theory of random graphs, but we significantly extend the conventional random graph model, as in its original definition it
would be too sterile to capture realistic wireless networks. A closed form expression for the network connectivity was derived
by an equilateral hexagon topology introduced from the minimum set covering problem. We also analyzed the effect of boundary
nodes on the connectivity of an infinitely and a finitely large network. Through a combination of mathematical proof and simulations,
we have shown that our result provides a robust performance in wireless networks. 相似文献
19.
针对传统物体识别算法中只依赖于视觉特征进行识别的单一性缺陷,提出了一种结合先验关系的物体识别算法。在训练阶段,通过图模型结构化表示先验关系,分别构建了图像-图像、语义-语义两个子图以及两子图之间的联系,利用该图模型建立随机游走模型;在识别阶段,建立待识别图像与随机游走模型中的图像节点和语义节点的关系,在该概率模型上进行随机游走,将随机游走的结果作为物体识别的结果。实验结果证明了结合先验关系的物体识别算法的有效性;提出的物体识别算法具有较强的识别性能。 相似文献
20.
在图结构数据上开展推理计算是一项重大的任务,该任务的主要挑战是如何表示图结构知识使机器可以快速理解并利用图数据。对比现有表示学习模型发现,基于随机游走方法的表示学习模型容易忽略属性对节点关联关系的特殊作用,因此提出一种基于节点邻接关系与属性关联关系的混合随机游走方法。首先通过邻接节点间的共同属性分布计算属性权重,并获取节点到每个属性的采样概率;然后分别从邻接节点与含有共有属性的非邻接节点中提取网络信息;最后构建基于节点-属性二部图的网络表示学习模型,并通过上述采样序列学习得到节点向量表达。在Flickr、BlogCatalog、Cora公开数据集上,用所提模型得到的节点向量表达进行节点分类的Micro-F1平均准确率为89.38%,比GraphRNA(Graph Recurrent Networks with Attributed random walks)高出了2.02个百分点,比经典工作DeepWalk高出了21.12个百分点;同时,对比不同随机游走方法发现,提高对节点关联有促进作用的属性的采样概率可以增加采样序列所含信息。 相似文献