无线传感器网络内部恶意节点识别关键技术研究

After nodes are captured and cracked in wireless sensor networks, they become malicious nodes to result in internal attacks. The existing cryptographic security machine cannot be guard against effectively. Our research will focus on the key technology of the malicious node identification in wireless sensor network, in order to resolve the bottlenecks of the high energy consumption to identify malicious nodes, the lack of precision, lack of generality and adaptability. In the collection and monitor the behaviors of nodes, we focus on the lightweight monitoring mechanism of node behavior, in order to further reduce the energy consumption of monitoring mechanism, we can select the suitable monitoring node and set the frequency of monitoring; in the evaluation of node behavior, the abnormal behavior is marked and natural fault is filtered. The internal malicious nodes are precisely identified by the evaluation results. In the research of the internal malicious node identification, we focus on the research of the dynamic evaluation model based on node behavior and establishing the dynamic evaluation of node behavior based on trustworthy reputation model. The internal malicious nodes are identified according to the evaluation results. In the research of the internal malicious node identification under a variety of environments, we design the dynamic evaluation model based on multidimensional decision attributes to evaluate the dynamic behavior of nodes, which can effectively guarantee the complexity and uncertainty in the evaluation of node behavior relationship. In this project, the research and implementation, the identification and defense of internal malicious nodes, is a useful complement to the traditional security mechanism.

面对无线传感器网络节点被捕获、破解，成为恶意节点导致的内部攻击，现有的基于密码学的安全机制不能有效防范。本项目研究无线传感器网络中内部恶意节点识别的关键技术，旨在解决无线传感器网络内部恶意节点识别存在的能耗高、结果精确度不足、缺乏通用性与自适应性等问题。在节点行为的收集与监测方面，研究轻量级的节点行为监测机制，通过监测节点的选择与频率设置，进一步减少监测机制中的能耗；在节点行为评估方面，对网络中异常行为标记与自然故障过滤，有效提升恶意节点识别的精确度；在内部恶意节点识别研究中，研究基于信任信誉模型的通用的节点行为动态评估模型，根据行为评估结果识别内部恶意节点。在多种环境下的内部恶意节点识别中，研究基于多维决策属性的动态评估模型，通过多维决策属性进行节点行为动态评估，实现节点行为关系中的复杂性与不确定性评估。通过本项目的研究可实现内部恶意节点的识别，是传统安全机制的有益补充。

面对无线传感器网络节点被捕获、破解，成为恶意节点导致的内部攻击，现有的基于密码学的安全机制不能有效防范。本项目研究无线传感器网络中内部恶意节点识别的关键技术，提出了基于滑动窗口的动态信任模型、基于风险感知的信任评估模型、多维决策属性的动态信任评估和一种基于信任机制的无线传感器网络安全数据融合算法等，所提出的方案在于解决无线传感器网络信任评估的动态性、实时性以及通用性，有效根据复杂环境实现内部恶意节点的识别，确保无线传感器网络应用的安全性。除此之外，本项目资助期间对其它研究方向产生了一些相关研究成果，主要包括：无线传感器网络高能效数据收集技术研究、云计算中的查询结果完整性验证技术研究、三维模型可逆信息隐藏技术研究和基于属性加权和聚类的决策树结点划分方法研究。本项目资助期间以第一批注共发表论文15篇，其中，SCI检索7篇，EI检索8篇；其中，期刊论文11篇，国际会议论文4篇，出版学术专著一部。在获得的知识产权方面，已授权的实用新型3项，公开的发明专利一项，软件著作权9项。在人才培养方面，共培养4名博士研究生顺利毕业获得博士学位。