Thwarting Wi-Fi Side-Channel Analysis through Traffic Demultiplexing

Wi-Fi网络已得到广泛部署。在Wi-Fi网络中，即使流量被加密，无线链路上的侧信道信息泄漏也是不可避免的，因为流量的特征是可以通过流量分析来识别的。作为一个序列，侧通道信息泄露在各种在线应用中都有报道。然而，Wi-Fi网络中仍然严重缺乏有效的解决方案来缓解信息泄漏。 在提出的研究中，我们将提出一种新的设计，称为流量解复用，以实现对流量分析的有效和高效防御。我们的想法是，在通过Wi-Fi通道传输数据包时，将Wi-Fi流量划分为多个片段，每个片段由数据流中的数据包子集组成。这可防止攻击者重新组合代码段以重建原始流的流量特征。我们的设计旨在提供针对流量分析的有效防御，而不会明显降低网络性能。为此，我们将开发一种MAC层流量划分方法，该方法包括MAC层虚拟化和多个虚拟MAC接口上的动态分组调度。以这种方式，单个虚拟接口的业务特性变得与原始流的业务特性显著不同，并且看起来独立于由其他接口产生的业务特性，从而为窃听者的业务分析增加了足够的挑战。 该研究将极大地改善传统的无线链路流量分析防御方法。它将推进今天的应用程序开发实践，并考虑到隐私问题，并将增强未来的无线通信标准。

Wi-Fi networks have been widely deployed. In Wi-Fi networks, even if the traffic is encrypted, the side-channel information leak over wireless links is inevitable, because the characteristics of traffic flows are identifiable through traffic analysis. As a sequence, side-channel information leaks have been reported in various online applications. However, a serious lack of efficient and effective solutions to mitigate information leaks persists in the Wi-Fi networks. In the proposed research, we will present a novel design, called traffic demultiplexing to achieve both effective and efficient defense against traffic analysis. Our idea is to partition a Wi-Fi traffic flow into snippets, each composed of a subset of the packets in the flow, when transmitting these packets through the Wi-Fi channel. This prevents adversaries from re-assembling the snippets to reconstruct the traffic characteristics of the original flow. Our design aims at providing an effective defense against traffic analysis without noticeable network performance degradation. To this end, we will develope a MAC layer traffic partition approach that includes MAC layer virtualization and dynamic packet scheduling over multiple virtual MAC interfaces. In this way, the traffic characteristics of a single virtual interface become significantly different from those of the original flow, and appear independent from those produced by other interfaces, thereby add enough challenges for traffic analysis by the eavesdropper. The proposed research will significantly improve the traditional defense against traffic analysis over the wireless links. It will advance today's application development practice with privacy concerns and will enhance future wireless communication standards.