基于侧信道可信融合的无线网络感知理论与关键技术研究

As the urgent needs of mobile computing in a novel array of of human-computer interaction, indoor navigation, trajectory tracking and medicine monitoring, the ubiquitous wireless networks and smart mobile devices-based networked sensing technologies have garnered significant attention recently in academia and industry. Existing RF signal backscatter-based human behavior recognition or tracking technologies are focused on modeling the channel change feature to improve accuracy or robustness of the system. However, utilizing this kind of side-channel vector for building networked sensing system is easy to leak sensitive information. The proposal attacks this issue how to exploiting online physical layer information confusion effective to prevent illegal user using perception equipment hacking without effecting original wireless communications system throughput and data quality. First of all, designing confusion and synthesis mechanism on wireless signal physical layer information cause by human behavior and environment changes; secondly, providing constraints conditions of side channel trustworthy fusion and mathematics model based on preventing SISO\MIMO system of side channel attack; thirdly, proposing a cross-spatial sensing and scale adaptive method without the constraints of wireless devices deployment; and finally, implementing and evaluating a trustworthy device-free passive indoor positioning and activity recognition system via the commercial wireless equipment and software radio platform.

随着创新人机交互、室内定位、轨迹追踪、医疗监护等移动计算的迫切需求，基于泛在无线网络和智能移动设备的网络感知技术受到学术界和工业界广泛关注。现有基于射频信号反向散射的无设备人体行为识别或轨迹追踪技术大都聚焦于使用获取信道变化特征建模提高计算精度或系统健壮性，然而利用这类侧信道矢量建立的网络感知系统存在天然的敏感信息泄露。项目围绕如何在不影响原有无线通信系统吞吐和数据质量的前提下，利用物理层信息有效在线混淆达到防止非法用户利用感知设备窃听这一主题展开研究。首先，合理设计无线信号物理层信息对人体行为和环境变化的特征建模和混淆合成机制；其次，给出基于SISO\MIMO系统的侧信道攻击防护的侧信道可信融合约束条件和数学模型；再次，提出一个不受无线设备部署约束的跨空间感知尺度自适应方法；最后，通过商用无线设备和软件无线电平台实现和评价一个室内无设备被动定位和动作识别可信感知系统。

利用泛在的无线网络设备收发端形成的射频信号空间对影响其反射、衍射和折射链路的人体及其行为进行无接触感知存在侧信道攻击风险。项目在分析现有侧信道攻击防护和无设备被动感知技术威胁模型基础上，提出并验证逻辑层保真而物理层失真的中继混淆方法，突破现有网络感知技术受感知设备信号部署空间约束的限制，提出跨空间感知尺度自适应的信号去噪与特征提取方法，使用户在不同位置不受感知设备与人体相对位置约束的被动感知具备技术可信；提出并验证基于侧信道矢量可信融合的高精度无设备被动感知方法；提出并验证反向散射信号侧信道矢量的多维度特征提取与建模方法，最后对以上理论成果、方法和技术进行实验验证，通过商用无线网络设备和软件无线电平台实现室内可信无设备被动定位和动作识别感知系统。这些理论和技术在链路层和信号层对无线网络感知空间安全进行了全面探究，可以推广到室内无线网络应用的各个场景。