Ad-Hoc网络中信任评估模型研究

发布时间：2018-10-15 15:19

【摘要】：由于具有自组织、无中心、节点资源有限、拓扑动态变化、无线信道完全开放等特征,Ad-Hoc网络容易遭受到各种恶意节点的攻击,如不合作、恶意推荐、合谋攻击等形式。Ad-Hoc网络的安全问题直接影响了网络的健壮性和网络寿命。信任模型是解决当前Ad-Hoc网络安全问题的重要方法,而如何能够精确地描述节点间的信任关系则成了任何信任管理模型都要面临的首要问题。本文针对Ad-Hoc网络节点间信任关系管理中的两个关键环节行为监视和信任评估进行了研究。在Ad-Hoc网络中节点通常会因为自身各种资源有限,导致其成为自私节点或恶意策略节点。如果自组织网络中长期存在大量的自私节点,将严重打击其他优秀节点提供可靠服务的积极性,移动自组织网络最终将难以为继。而恶意策略节点有选择有策略地提供不可靠服务,也会影响Ad-Hoc网络的服务质量。因此,需要设计一种有效机制使移动自组网中的节点积极提供可靠的服务,提高网络的业务能力是解决问题的关键。由于Ad-Hoc网络缺乏集中式服务器,节点的信任不能由第三方来计算。同时由于网络拓扑的变化或无线冲突的发生,用于信任评估的证据抽样空间可能会不完全或不可靠,从而影响信任评估有效性。因此,针对以上问题,本论文围绕Ad-Hoc网络中的信任评估模型进行研究,主要成果如下： (1)提出在直接信任的更新过程中,根据信任向量相似度的大小来确定加权系数,利用激励机制来惩罚恶意节点,奖励忠诚节点,做到惩罚力度大于奖赏力度。该措施可以有效抑制恶意节点快速提升自己的信任,迫使恶意节点能够遵循移动自组织网络初衷,更好的为其他节点提供服务。通过精确描述节点间的直接信任关系,提出了基于相关度和激励机制的有利于提高网络中节点交易成功率的直接信任合成规则,用来增强网络的安全性和健壮性。 (2)实时监测推荐节点行为、及时发现恶意节点、为推荐节点建立信任合成规则在面对不安全的网络环境时可以提高网络安全性与鲁棒性。针对在信任推荐过程中,恶意节点可能会发起诽谤攻击或共谋攻击来提供虚假信任,虚假信任与其他信任之间的证据冲突导致信任合成与直觉相悖,产生不合理结果的问题,提出了基于集合的推荐信任过滤管理机制来监测推荐节点的行为,利用D-S理论中带有折扣系数的合成规则来解决该问题。
[Abstract]:Because of its characteristics of self-organization, no center, limited node resources, dynamic topology change and completely open wireless channel, Ad-Hoc network is vulnerable to various malicious node attacks, such as non-cooperation, malicious recommendation, etc. In the form of collusion attack, the security problem of Ad-Hoc network has a direct impact on the robustness and lifetime of the network. Trust model is an important method to solve the current Ad-Hoc network security problems, and how to accurately describe the trust relationship between nodes has become the most important problem for any trust management model. In this paper, the behavior monitoring and trust evaluation of two key links in the management of trust relationship between Ad-Hoc network nodes are studied. In Ad-Hoc networks, nodes usually become selfish nodes or malicious policy nodes because of their limited resources. If there are a large number of selfish nodes in the Ad Hoc network for a long time, the enthusiasm of other excellent nodes to provide reliable services will be severely dampened, and the mobile ad hoc network will ultimately be difficult to sustain. The malicious policy nodes provide unreliable services selectively and strategically, which will also affect the quality of service of Ad-Hoc networks. Therefore, it is necessary to design an effective mechanism to enable the nodes in Manet to provide reliable services actively, and the key to solve the problem is to improve the service capability of the network. Due to the lack of centralized servers in Ad-Hoc networks, the trust of nodes cannot be calculated by third parties. At the same time, because of the change of network topology or the occurrence of wireless conflict, the evidence sampling space used for trust assessment may be incomplete or unreliable, thus affecting the validity of trust assessment. Therefore, this paper focuses on the trust assessment model in Ad-Hoc networks. The main results are as follows: (1) in the process of direct trust updating, The weighted coefficient is determined according to the similarity of the trust vector, and the incentive mechanism is used to punish the malicious node and reward the loyal node, so that the punishment is greater than the reward. This measure can effectively restrain malicious nodes from rapidly enhancing their trust and force malicious nodes to follow the original intention of mobile ad hoc networks and provide better services to other nodes. By accurately describing the direct trust relationship between nodes, a direct trust composition rule based on correlation and incentive mechanism is proposed to improve the transaction success rate of nodes in the network. It is used to enhance the security and robustness of the network. (2) Real-time monitoring of the behavior of recommendation nodes, timely detection of malicious nodes, Establishing trust composition rules for recommendation nodes can improve network security and robustness in the face of insecure network environment. In the process of trust recommendation, malicious nodes may launch libel attacks or collusion attacks to provide false trust. The conflict of evidence between false trust and other trust results in the combination of trust against intuition and unreasonable results. A recommendation trust filtering management mechanism based on set is proposed to monitor the behavior of recommendation nodes. The synthesis rule with discount coefficients in D-S theory is used to solve this problem.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08;TN929.5

【参考文献】