基于嵌入式的P2P流量监控系统的设计及实现

P2P业务消耗大量的网络带宽资源,使互联网传统业务的服务质量下降,针对这个问题,设计了一种基于嵌入式技术的P2P流量监控系统.该嵌入式系统采用深层数据包检测(deep packet inspection,DPI)的流量检测技术,使用旁路干扰的控制方式,并具有扩展到对新的P2P业务流量监控的功能.测试结果表明该嵌入式系统检测和控制P2P业务流量的效果良好,扩展性较强.     

应用感知的研究

本文主要是对DPI等应用感知技术进行研究分析,并在此基础上提出了一个新的应用层流量识别模型,新模型除了能够实现对HTTP、FTP、SMTP、POP3等传统应用以及BitTorrent、eDonkey、Skype、Gnutella等P2P应用的分类识别,还具备一定的学习能力,即能够学习并识别一些未知P2P协议。     

Deep Packet Inspection: Shaping the Internet and the Implications on Privacy and Security

ABSTRACT

In recent debates on the issues of privacy and the Internet, Deep Packet Inspection (DPI) has been grossly oversimplified as a technology that is primarily harmful to consumers. Much of the commentary has been based on a poor understanding of what DPI is and how DPI works. All too often the debate over DPI is focused on unrelated issues such as free speech and censorship, which are largely political and not technological issues. DPI usage has been described as a violation of consumer privacy when the reality is far more complex and nuanced. What has been missing is a broader discussion on the full nature and application of DPI technology; DPI enables a wide range of applications, most of which are not only positive but also essential to the survival of the Internet. This paper will explain how DPI technology works and explore practical applications of its use. DPI will be an important tool to control economies of scale with the explosive growth of the Internet. While there will always be privacy concerns over the intrusive nature of DPI technology, this worry will be overshadowed by the security and control it allows.     

网络用户的监控与管理

随着互联网普及率稳步提升,网络反映和协调社会功能的作用不断得到拓展和延伸。但是,网络也成为不法分子传播淫秽暴力信息、进行网络诈骗等不法活动、随意谩骂进行人身攻击的空间。这给网络案件的侦破带来难度,容易发生名誉权侵权纠纷。如何对网络用户行为进行监控和追查成为了网络管理者的一项重要任务,为此聊城大学网络信息中心引进了RG-eLog锐捷日志系统来更好的进行网络日志管理。高校是网络用户相对集中的地方,而高校学生是网络用户中比较活跃的群体,为此聊城大学网络信息中心在对网络用户名命名上借鉴了网络实名制的思想,为充分RG-eLog锐捷日志系统建立了一个必要基础。     

校园网P2P流量检测研究

随着P2P应用的飞速发展,监控P2P已经不可避免,当前的检测技术中,DPI检测的主要缺点是需要对网络上的每一个数据包进行模式匹配,时间复杂度相当大;流检的主要缺点是识别精度不高,而且要缓存网络上大量的数据包,需要较大的空间复杂度,本文提出一种经过双层过滤的P2P流量检测模型,旨在结合两者的优点,避免两者重复产生的时间和空间复杂度。     

Crystal engineering of lactose using electrospray technology: carrier for pulmonary drug delivery

Context: Dry powder inhalers (DPIs) consisting of a powder mixture containing coarse carrier particles (generally lactose) and micronized drug particles are used for lung drug delivery. The effective drug delivery to the lungs depends on size and shape of carrier particles. Thus, various methods have been proposed for engineering lactose particles to enhance drug delivery to lungs.

Objective: The objective of current work was to assess suitability of electrospray technology toward crystal engineering of lactose. Further, utility of the prepared lactose particles as a carrier in DPI was evaluated.

Materials and methods: Saturated lactose solutions were electrosprayed to obtain electrosprayed lactose (EL) particles. The polymorphic form of EL was determined using Fourier transform infrared spectroscopy, powder X-ray diffractometry, and differential scanning calorimetry. In addition, morphological, surface textural, and flow properties of EL were determined using scanning electron microscopy and Carr’s index, respectively. The aerosolization properties of EL were determined using twin-stage impinger and compared with commercial lactose particles [Respitose^® (SV003, Goch, Germany)] used in DPI formulations.

Results and discussion: EL was found to contain both isomers (α and β) of lactose having flow properties comparable to Respitose^® (SV003). In addition, the aerosolization properties of EL were found to be significantly improved when compared to Respitose^® (SV003) which could be attributed to morphological (high elongation ratio) and surface characteristic (smooth surface) alterations induced by electrospray technology.

Conclusion: Electrospray technology can serve as an alternative technique for continuous manufacturing of engineered lactose particles which can be used as a carrier in DPI formulations.     

面向移动应用识别的结构化特征提取方法

针对移动应用流量监控及行为分析等需要,为有效识别移动网络流量所属的应用,提出一种超文本传输协议（HTTP）流结构化特征提取方法。采取一款自研的基于虚拟专用网络（VPN）的流量采集工具获取研究数据,该工具能够精确标识每一条数据流归属的应用。在特征提取阶段,不预先设计特征构成,通过流聚类、获取最长公共子序列、字符替换得到应用HTTP流的结构化特征。从42种应用的117 772条HTTP流中提取特征,并对测试集的50 387条HTTP流进行识别,所提方法的平均准确率达99%,平均查全率为90.63%,单个应用最大误报率为0.52%。实验结果表明,该结构化特征提取方法能够有效识别移动应用的流量。     

无源相干定位系统直达波干扰对消研究

无源相干定位系统中的主要问题之一就是接收机接收到的目标信号回波中的直达波干扰。传统的解决办法是采用自适应天线零陷对准直达波干扰,但广播信号的无源相干定位系统的天线零陷不够低。因此,文中提出了采用双通道对消的办法来抑制直达波,给出了所有的数学推导以及推导过程中的问题,最后给出了实验对消的结果。     

基于DPI的用户兴趣实时分类

近年来,随着大数据发展热潮兴起,电信运营商能获得的DPI数据规模越来越大。如何对如此海量的DPI数据进行实时的挖掘和分类是运营商研究的热点。设计了一种基于混合n-gram特征的URL分类算法和基于Doc2Vec模型的文本分类算法相结合的DPI数据分类方法,大大提高了DPI数据的分类速度和分类准确率,为电信运营商进行精准营销提供了良好的基础。     

一种联合DPI和DFI的网络流量检测方法

提出一种以深度包检测(DPI)技术为主、深度流检测(DFI)技术为辅的网络流量检测方法。基于MPC8572网络处理器的模式匹配引擎模块,利用DPI实现细粒度检测,对于DPI的误识别情况,通过DFI进行鉴别并提示重新检测,以达到纠错目的。实验结果表明,联合方法具有检错和纠错功能,且能提高网络流量检测的准确率。