混合结构机会网络链路预测技术研究

发布时间：2018-05-07 12:39

本文选题：机会网络 + 机会路由　；参考：《东南大学》2015年硕士论文

【摘要】：大量具备短距离无线通信能力的智能设备的普及推动了机会网络的研究和应用。但在实际应用环境中,由于节点移动、网络稀疏或信号衰减等各种原因,机会网络中的链路动态性较高,导致消息投递率成功低、投递时延大等问题,制约了机会网络的实用性。在机会网络中部署具有较大缓存空间、位置固定或者移动的辅助节点,形成混合结构机会网络,可以降低网络的动态性,提高网络传输性能。混合结构机会网络符合诸如校园等社区网络、路线固定的公交车网络、具有路边节点的车载网络等应用形式。本论文主要研究基础设施辅助节点对网络动态性,链路变化可预测性的影响以及对消息投递性能的改进作用,内容如下：(1)研究混合结构网络统计特性。为了研究混合结构机会网络中辅助节点的数量和密度等参数对网络的影响,本论文通过对实际数据集进行统计和分析的方式进行一系列的实验,总结了网络动态性、链路属性的变化。实验结果表明：节点相遇时间间隔分布函数曲线呈现“重尾”特征；辅助节点的密度和部署方法对提升机会网络的传输性能有较大影响：仅仅少量正确部署的辅助节点就能显著提升网络性能,而高密度的辅助节点带来的边际效用递减。(2)提出混合结构机会网络中的链路预测组合算法。该算法对相遇频繁且呈周期性的节点对采用基于周期模式挖掘的链路预测算法,对相遇频繁但非周期性的节点对采用基于决策树的机器学习链路预测算法,对相遇不频繁的节点对采用基于相似性的链路预测算法。在实际数据集中进行的实验结果表明,该组合算法的预测效果要优于其他单一算法,且该算法能够对更多的节点对实现预测。(3)研究基于预测的混合结构机会网络消息投递。结合(1)的结论,提出了Prophet-F和基础设施节点的部署方案;结合(2)的组合链路预测算法,提出基于PDR的路由协议；并在实际数据集的基础上进行模拟实验。结果表明将基础设施部署在热点位置网络性能最好,且在确定部署方案下Prophet-F较网络性能提高15%～20%；基于PDR的路由协议性能优于包括Prophet-F在内的其他路由。论文成果能够提高机会网络的实用性,可以改善社区机会网络、车载自组织网络等机会网络的传输性能。
[Abstract]:The popularity of a large number of smart devices with short-range wireless communication capabilities has promoted the research and application of opportunistic networks. However, in the practical application environment, due to various reasons, such as node movement, network sparsity or signal attenuation, the link dynamics in the opportunistic network is relatively high, which leads to the low message delivery rate and the large delivery delay. It restricts the practicability of the opportunity network. The deployment of auxiliary nodes with large buffer space, fixed position or moving position in opportunistic networks to form a hybrid opportunistic network can reduce the dynamic nature of the network and improve the transmission performance of the network. The hybrid structure opportunity network accords with community network such as campus, bus network with fixed route, vehicular network with roadside node and so on. In this paper, the influence of infrastructure auxiliary nodes on network dynamics, the predictability of link changes and the improvement of message delivery performance are studied. The main contents are as follows: 1) the statistical characteristics of hybrid networks are studied. In order to study the influence of the parameters such as the number and density of auxiliary nodes on the network, this paper makes a series of experiments through statistics and analysis of the actual data sets, and summarizes the dynamics of the network. Changes in link attributes. The experimental results show that the distribution function curve of time interval between nodes meets with "heavy-tailed" feature. The density and deployment method of auxiliary nodes have great influence on the transmission performance of opportunistic networks: only a small number of properly deployed auxiliary nodes can significantly improve the network performance. A link prediction combination algorithm is proposed for hybrid opportunistic networks with reduced marginal utility brought by high density auxiliary nodes. The algorithm adopts link prediction algorithm based on periodic pattern mining for frequent and periodic node pairs, and machine learning link prediction algorithm based on decision tree for frequent but aperiodic node pairs. A link prediction algorithm based on similarity is used for node pairs that meet infrequently. The experimental results in the actual data set show that the prediction effect of the combined algorithm is better than that of the other single algorithms, and the algorithm can be used to realize the prediction of more node pairs. In combination with the conclusion of No.1), the deployment scheme of Prophet-F and infrastructure nodes is proposed, and the routing protocol based on PDR is proposed in combination with the combined link prediction algorithm of X2), and the simulation experiments are carried out on the basis of the actual data set. The results show that the performance of the network is the best when the infrastructure is deployed in the hot spot, and the performance of Prophet-F is better than that of the network under certain deployment scheme, and the performance of the routing protocol based on PDR is better than that of other routes, including Prophet-F. The results of this paper can improve the practicability of the opportunity network and improve the transmission performance of the opportunity network such as the community opportunity network and the vehicle-mounted ad hoc network.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.02

【相似文献】