移动自组网中非完全信息节点风险评估

Incomplete Information Nodes Risk Assessment in Mobile Ad Hoc Networks

移动自组网是一种无中心、自组织以及多跳的无线网络,能否有效可靠地对通信节点进行风险评估是保障节点高质量通信的重要因素.当前基于可信度或是信誉的评估方法对评估样本都有特殊的要求,对评估对象提出种种假设,无法辨别间接样本的恶意推荐问题,在实际移动自组网中各个节点尤其是恶意节点对外暴漏的信息往往偏少,样本数量有限,无法明确判断其状态,大部分是非完全信息,从而导致评估结果缺乏客观性与可靠性.文中提出采用灰色系统理论描述通信节点非完全信息状态,根据灰类白化以及灰色聚类思想进行节点风险评估.方案针对自组网络通信需求对节点行为进行多关键属性划分,采用味集群方法收集原始样本,避免恶意推荐;将难以用数值精确刻画的关键属性以白化权函数量化,引入灰聚类的概念和计算方法,将实体的通信风险水平定义为实体关键属性值序列针对各评估灰类的聚类评估值,从而得到该实体通信风险水平的相对参考值.分析与实际计算表明该方法是一种适合移动自组网中非完全信息节点风险评估的有效方法.

Mobile Ad hoe Networks are distributed self-organized and multi-hop networks, the efficient networks nodes risk assessment is the key factor for high quality communication. Most assessment methods based on reliability or reputation have particular requirement, suppose some conditions to the samples, and can＇t identify the vicious recommendation. However, in true Ad hoc networks, the exposed information of some nodes is little, which is more serious to some vicious nodes, the samples are so little that the clear state can＇t be identified and most are incomplete information, which result in that the assessment is not objective and credible. According to these problems, a Ad hoc networks risk assessment method based on grey theory is proposed. In this scheme, the communication nodes incomplete information state is described as grey league, the whitening function and grey clustering idea are used to do the assessment. The key attributes of nodes are marked off, original data is collected using taste concourse method to avoid vicious recommendation, the evaluated nodes are regarded as clustering entities, and the risk grade point value of their key attributes are given by clustering set assessment value, the analysis and computing shows this scheme is a efficient incomplete information nodes risk assessment method in mobile Ad hoc networks.