【摘要】 随着网络的发展,网络地址转换(NAT)技术的应用越来越普及,缓解了网络IP地址枯竭的问题,然而也带来了一定的安全隐患。因此,有效地识别NAT变得十分有必要。论文提出一种基于网络流量的NAT识别技术,引入投影寻踪技术,通过网络流量特征对NAT进行识别。使用基于量子粒子群的投影寻踪算法,将量子粒子群的全局搜索能力与投影寻踪对高维数据的降维能力的结合。核心是获取网络中IP地址设备的流量特征数据,通过算法把高维数据投影到低维子空间上,从而将NAT设备与普通设备分隔开,实现NAT的识别。更多还原

【Abstract】 With the development of network,network address translation(NAT)technology is more and more widly applied,which relieves the problem of network IP address exhaustion,and brings some potential risk.Therefore,the effective identification of NAT become very necessary.This paper proposes a NAT identification technology based on network flow,introducing the projection pursuit technique,analysising on the characteristics of network flow to identify NAT.Using projection pursuit based on quantum particle swarm algorithm,combing the quantum particle swarm global search ability and the high-dimensional data dimension reduction ability of the projection pursuity.Core is to obtain the feature data of the network equipment flow,using the algorithm to project the high-dimensional data to low-dimensional subspace,which will separate the pervasive devices and the NAT devices,to achieve the identification of NAT.更多还原