URL: https://infocom2020.ieee-infocom.org/accepted-paper-list-main-conference
选读网络流量分类领域的论文。
- Autonomous Unknown-Application Filtering and Labeling for DL-based Traffic Classifier Update
Jielun Zhang, Fuhao Li, Feng Ye and Hongyu Wu (University of Dayton, USA)
论文URL:https://arxiv.org/pdf/2002.06359.pdf
主要内容：
传统的基于机器学习的流量分类模型，都是在close-world里面做的实验，这些模型不能分类未知的类别。作者提出自主机器学习框架来更新流量分类器。这个框架包括一个基于深度学习的分类器，一个自学习的鉴别器，一个自标注的模型。鉴别器和自标注器可以生成新的数据集共分类器重新学习。这玩意先识别未知的流量，然后给未知流量自动打标签。
这个玩意和陈同学的工作特别相似。
这是一篇水文。 - Exposing the Fingerprint: Dissecting the Impact of the Wireless Channel on Radio Fingerprinting
Amani Al-Shawabka, Francesco Restuccia, Salvatore D’Oro, Tong Jian, Bruno Costa Rendon, Nasim Soltani, Jennifer Dy, Stratis Ioannidis, Kaushik Chowdhury and Tommaso Melodia (Northeastern University, USA)
论文地址：https://ece.northeastern.edu/fac-ece/ioannidis/static/pdf/2020/AlShawabka2020Exposing.pdf
主要内容：
大规模的测量了CNN网络通过无线电指纹（Radio Fingerprinting）来识别无线设备的能力。他们的输入是频谱图。这个应该是物理攻击了。 - MagPrint: Deep Learning Based User Fingerprinting Using Electromagnetic Signals
Lanqing Yang, Yi-Chao Chen, Hao Pan, Dian Ding, Guangtao Xue, Linghe Kong, Jiadi Yu and Minglu Li (Shanghai Jiao Tong University, China)
论文连接：http://www.cs.sjtu.edu.cn/~yichao/pmwiki/assets/publications/Infocom20_Yang.pdf
主要内容：
这篇论文提出用户的电磁指纹用于对用户进行连续的身份认证。
电磁指纹的定义：
In this study, we proposed to use electromagnetic (EM)
signals as a side-channel for fingerprinting users. The intensity
of EM signals generated from a mobile device reflects the
computational intensity of the device. For example, a heavy
computational load tends to push up power consumption by
the CPU, which leads to an increase in EM induction. Furthermore, users vary considerably in the way that they use their
devices in terms of key hold duration, typing interval, cursor
speed, pause and click time, etc [5]–[7]. The instructions
associated with these actions vary the internal operations of
the device and are thus reflected in the EM signals.
如何采集电磁指纹？
使用一个电磁传感器，采集的就是普通的磁场信号。也可以是设备内置的电磁传感器。
与传统的基于手纹和前置摄像头相比有什么优势？
传统的指纹采集工具不能一直对用户进行认证，而前置摄像头又会对用户造成比较明显的干扰。