分布式移动社交网络中移动模型和路由算法的研究

发布时间：2018-05-29 15:26

本文选题：分布式移动社交网络 + 家　；参考：《北京交通大学》2017年硕士论文

【摘要】：互联网产业技术的迅速发展,智能移动终端设备的大规模普及,3G、4G通信技术的广泛应用,促进了社交网络和移动终端的自然结合,分布式移动社交网络应运而生。分布式移动社交网络就是将人的移动性、社会性与无线通信技术相结合的一种特殊的时延容忍网络。但在该网络中,由于节点的移动性和间歇性连接,给信息传输带来极大挑战。为了促进分布式移动社交网络的应用与发展,设计能反映人们真实移动规律的移动模型和高效的路由算法是特别重要的。这也是本文的主要研究内容。为了设计一个能反映人们真实移动规律的移动模型,本文首先对Dartmouth/campus数据集进行了统计分析,得出在现实生活中,每个人在大多数的时间内都会在几个固定的地点之间相互转移,而且较长时间内这些地点是稳定的。基于此特性,本文延伸了家的概念,家不再仅指人们居住的地点,而是人们经常访问的地点。然后设计了一个基于家的移动模型—HBMM。在HBMM中,首先收集一段时间内人们的移动轨迹,然后根据收集到的数据来确定人们家的集合,计算在家中的停留时间和家之间的转移概率,最后形成马尔可夫状态转移图。状态转移图中人们会以较大的概率在家的集合内部相互转移,以较小的概率转移到其他不经常访问的区域。通过该状态转移图就可以预测人们在很长一段时间内的移动轨迹。实验结果表明此移动模型能够较为准确地再现人们实际的移动情况。为了提高分布式移动社交网络中消息的转发效率,在移动模型研究的基础上,设计了三种基于家的多中继路由选择算法:HBMRZ、HBMRA和HBMR。HBMRZ中假定消息在分发阶段不知道目的节点家的位置,为了提高消息的传输成功率,设计了该情况下的最优化中继选择策略,通过该策略,可以使所有中继节点家的覆盖范围最广,从而提高消息的传输成功率。HBMRA中假定消息分发阶段知道目的节点家的信息,该情况下我们就会选择那些与消息目的节点家的集合相似度高的节点作为消息的中继节点。HBMR是一种实际情况,每个节点都会建立一张类似于路由表的关于节点家的信息表,节点相遇时,会彼此交换更新信息表。当节点进行消息转发时,就可以通过信息表获取到消息目的节点家的信息。通过仿真实验,也证明了这三种算法都是优于传统的多中继选择算法SprayWait和HS的,而且HBMRA的性能还要远远优于HBMRZ,HBMR的性能趋近于HBMRA。
[Abstract]:With the rapid development of Internet technology and the wide application of 3G 4G communication technology, the distributed mobile social network emerges as the times require. Distributed mobile social network is a special delay-tolerant network which combines mobility, sociality and wireless communication technology. However, due to the mobility and intermittent connection of nodes in the network, it brings great challenges to the transmission of information. In order to promote the application and development of distributed mobile social networks, it is very important to design mobile models and efficient routing algorithms that can reflect the real movement of people. This is also the main research content of this paper. In order to design a movement model that can reflect the real movement of people, this paper firstly makes a statistical analysis of the Dartmouth/campus data set, and concludes that in real life, everyone will transfer to each other in several fixed places most of the time. And these sites are stable for a long time. Based on this feature, this paper extends the concept of home, which not only refers to the place where people live, but also the place that people often visit. Then, a home-based mobile model, HBMMM-based, is designed. In HBMM, people's moving trajectories are collected for a period of time, then the collection of people's homes is determined according to the collected data, the residence time and the transition probability between homes are calculated, and finally a Markov state transition map is formed. In the state transition graph, people will transfer to other infrequently visited areas with a high probability of transferring to each other within the set at home, and to other infrequently visited areas with a lower probability. The state transition graph can predict the trajectory of people over a long period of time. The experimental results show that the model can accurately reproduce the actual movement of people. In order to improve the efficiency of message forwarding in distributed mobile social networks, based on the research of mobile model, three home-based multi-relay routing algorithms: HBMRZ HBMRA and HBMR.HBMRZ are designed, which assume that the message does not know the location of the destination node in the distribution phase. In order to improve the success rate of message transmission, the optimal relay selection strategy is designed in this case. Through this strategy, all relay nodes can have the widest coverage. In order to improve the success rate of message transmission. HBMRA assumes that the message distribution phase knows the information of the destination node, In this case, we will select those nodes that have a high similarity to the message destination family as the relay node of the message. HBMR is a practical situation, each node will establish a similar routing table about the home of the node information table. When nodes meet, they exchange updated information tables with each other. When the node forwards the message, the information of the destination node can be obtained through the information table. The simulation results show that these three algorithms are superior to the traditional multi-relay selection algorithms (SprayWait and HS), and the performance of HBMRA is much better than that of HBMR.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5;TP393.09

【参考文献】