机会社会网络中基于社交信息的数据投递机制研究

发布时间：2018-06-25 02:55

本文选题：机会社会网络 + 节点行为　；参考：《东南大学》2016年硕士论文

【摘要】：机会社会网络(Social Opportunistic Networks)是一种以人为主体的机会网络,具有节点移动接触频繁、延迟较高、网络间歇性连接、节点运算能力和缓存资源受限等特点。与普通机会网络不同的是,机会社会网络中移动节点主要由人的意志主导,因此其在移动和接触行为上具有明显的社会性。分析社会属性是研究机会社会网络中节点行为的主要方式之一,但是现有的大部分研究仅在社会属性对节点行为的影响上进行了定性分析,在量化评估其具体的影响程度方面工作较少,这为相关投递算法的设计和实际表现带来了一定的缺陷和误差。此外,多数基于社会属性的投递机制的研究容易忽略一些问题,例如依靠中心性所设计的投递算法在高负载情况下会导致局部网络拥塞、缓存溢出和资源消耗不均等负面影响,利用相似性进行投递的算法也会因孤僻节点而影响性能。本文将主要针对大量手持近距离通信设备组成的机会社会网络场景进行研究,考察在网络连通较差或网络部署成本昂贵的社交场合,仅依靠近距离通信设备提供数据传输服务的过程。研究内容主要分为节点社会行为模型、数据投递算法、TTL和缓存替换机制三个方面,具体如下：1)基于对节点的社会属性和Trace数据进行的分析,首先验证了社会属性相同对节点相遇接触产生的积极影响,然后提出了社会属性影响的相遇模型,对比随机相遇模型下的接触频率,量化估算不同社会属性对节点相遇产生的影响权重；2)基于社会属性对节点相遇产生的影响,首先提出了基于属性权重的投递算法,在投递过程中将消息投递给与消息目的节点在社会属性上更相似的节点；随后为了避免相似性投递算法中孤僻节点带来的负面影响,本文通过消息转发频率、相遇节点范围两个指标评估节点的活跃性(中心性),选择投递节点时绕过活跃性较低的节点；最后引入了协调因子,调整相似性与中心性在算法执行时占有的比例,以适用更多的场景；3)根据不同消息的平均投递延迟,提出基于属性权重的TTL分配策略,为不同的消息分配相应的过期时间,以加速冗余消息在网络中的消除；为了降低缓存溢出所带来的负面影响(比如投递率下降、网络资源浪费等),提出了属性价值最大化的替换算法(CMV),在缓存溢出时丢弃投副本数量最少、投递期望最低的消息。4)对本文所提出的投递算法、缓存替换策略等进行仿真实验,验证了其有效性,并将其和机会网络中常见的算法进行对比,验证了其在性能上的优势。
[Abstract]:Social opportunity Networks is an opportunity network with human being as the main body. It has the characteristics of frequent mobile contact, high delay, intermittent connection, limited computing power and cache resources. Different from the ordinary opportunity network, the mobile node in the opportunity social network is dominated by human will, so it has obvious sociality in mobility and contact behavior. The analysis of social attributes is one of the main ways to study node behavior in opportunity social networks, but most of the existing studies only make qualitative analysis on the impact of social attributes on node behavior. There is little work in quantitative evaluation of its specific impact degree, which brings some defects and errors to the design and actual performance of the related delivery algorithm. In addition, most researches on the mechanism of social attribute based delivery are easy to ignore some problems, for example, relying on the centrality of the delivery algorithm can lead to local network congestion, buffer overflow and unequal resource consumption in the case of high load. The algorithm based on similarity also affects the performance of isolated nodes. In this paper, we will focus on the opportunistic social network scenarios composed of a large number of handheld close-range communication devices, and examine social situations where network connectivity is poor or network deployment is expensive. The process of providing data transmission services only by close-range communication devices. The research is mainly divided into three aspects: node social behavior model, data delivery algorithm TTL and cache replacement mechanism, as follows: 1) based on the analysis of social attributes and trace data of nodes, First, the positive effects of the same social attributes on the node encounter are verified, and then the encounter model of the social attribute influence is proposed, and the contact frequency under the random encounter model is compared. Quantitative estimation of the impact weight of different social attributes on node encounter 2) based on the impact of social attributes on node encounter, a delivery algorithm based on attribute weight is proposed. In order to avoid the negative influence of the isolated nodes in the similarity delivery algorithm, this paper passes through the message forwarding frequency to deliver the message to the node that is more similar to the social attribute in the message destination node in the process of delivery, and then in order to avoid the negative influence of the isolated node in the similarity delivery algorithm, Two indexes in the range of encounter nodes are used to evaluate the activity (centrality) of the nodes, and to bypass the nodes with low activity when the nodes are delivered. Finally, the coordination factor is introduced to adjust the proportion of similarity and centrality in the execution of the algorithm. According to the average delivery delay of different messages, a TTL allocation strategy based on attribute weight is proposed to allocate the corresponding expiration time for different messages, so as to accelerate the elimination of redundant messages in the network. In order to reduce the negative effects of cache overflow (such as drop in delivery rate, waste of network resources, etc.), an attribute value maximization replacement algorithm (CMV) is proposed, in which the number of dropped copies is the least. The simulation experiments on the proposed delivery algorithm, cache replacement strategy and so on are carried out to verify its effectiveness, and compare it with the common algorithms in the opportunistic network to verify its performance advantages.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN929.5

【相似文献】