基于模糊理论的可信网络信任模型研究

发布时间：2018-05-03 19:33

本文选题：对等网络 + 模糊综合评判　；参考：《江西理工大学》2014年硕士论文

【摘要】：网络的规模呈几何级数增长，，网络中的各种应用也越来越普及，在人们生活中有着不可或缺的地位。与此同时，由于可信网络的自治性、动态性、匿名性，网络中的恶意节点的攻击行为对网络的稳定性和可用性造成了很大破坏，这些不安全因素也制约了各种网络应用的发展推广。信任模型的出现给上述问题的解决提供了新的途径。通过对国内外信任机制相关文献进行归纳研究，发现目前的信任模型大多注重信任计算值的准确性，往往忽视了信任本身的模糊性，本文针对这一问题，在综合国内外相关研究的基础上，对基于模糊理论的推荐信任模型进行研究，研究的主要内容包括：（1）P2P网络下基于模糊评判的推荐信任模型。出于提高P2P系统中交易安全性以及提高信任度量准确性的目的，提出一种基于模糊综合评判的推荐信任模型。为了充分体现节点间信任关系的模糊性、动态性，采用二层模糊综合评判方法来计算直接信任值，为了降低了人为赋值的误差，将交易金额和交易时间作为二层评判的权重。在推荐信任计算中引入节点交易密度函数和推荐共谋检测方法来计算推荐节点的推荐可信度。通过仿真实验验证了该模型可以有效抵御共谋推荐且对恶意节点有较高的检测率。（2）基于用户行为的模糊理论信任研究。本文以复杂网络环境为研究背景，运用模糊综合评判理论，分析研究了针对基于用户行为的声誉评估策略，在计算直接信任值的时候引入时间段的概念，对不同时间段的交易给予不同的权重处理，使得信任的累积具有实时性和动态性；在计算推荐信任值时加入“震荡因子”来抵抗震荡攻击，提高了信任机制的健壮性；对如何选择推荐路径进行探讨，给出了较为合理的推荐方法。最后进行系统仿真实验，通过在不同恶意节点攻击的情况下与Beth模型的相关对比验证了本模型的有效性和可用性。
[Abstract]:The scale of network is increasing in geometric series, and the application of network is becoming more and more popular, which plays an indispensable role in people's life. At the same time, because of the autonomy, dynamic and anonymity of trusted network, the attack behavior of malicious nodes in the network has caused great damage to the stability and availability of the network. These insecurity factors also restrict the development and promotion of various network applications. The emergence of trust model provides a new way to solve the above problems. Through the research on the related literatures of trust mechanism at home and abroad, it is found that most of the current trust models pay attention to the accuracy of trust calculation, and often ignore the fuzziness of trust itself. This paper aims at this problem. The recommendation trust model based on fuzzy theory is studied on the basis of synthesizing relevant research at home and abroad. The main contents of the research include: Recommendation trust model based on fuzzy judgment in P2P network. In order to improve the transaction security and the accuracy of trust in P2P system, a recommendation trust model based on fuzzy comprehensive evaluation is proposed. In order to fully reflect the fuzziness and dynamics of the trust relationship between nodes, the two-layer fuzzy comprehensive evaluation method is used to calculate the direct trust value. In order to reduce the error of artificial assignment, the transaction amount and transaction time are taken as the weight of the two-layer evaluation. The node transaction density function and the recommendation collusion detection method are introduced in the recommendation trust calculation to calculate the recommendation credibility of the recommendation node. The simulation results show that the model can resist collusion recommendation effectively and has a high detection rate for malicious nodes. Research on fuzzy theory trust based on user behavior. Based on the complex network environment and the fuzzy comprehensive evaluation theory, this paper analyzes and studies the reputation evaluation strategy based on user behavior, and introduces the concept of time period when calculating the direct trust value. Giving different weights to transactions in different time periods, making the accumulation of trust real-time and dynamic, adding "shock factor" to resist shock attack in the calculation of recommended trust value, improving the robustness of trust mechanism; How to choose the recommended path is discussed, and a more reasonable recommendation method is given. Finally, the system simulation experiments are carried out to verify the effectiveness and availability of the model by comparing it with the Beth model in the case of different malicious node attacks.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【参考文献】