基于信道状态信息的被动室内入侵检测理论与应用研究

Wireless passive indoor intrusion detection is based on passive indoor localization. It utilizes the rationale that different subjects have different influence on the transmitting of wireless signals. Wireless passive indoor intrusion detection is composed of two stages: human detection and identity authentication. Most current human detection and identity authentication techniques are based on strict assumptions, which results in weak universality and inapplicable in indoor intrusion detection. To overcome these problems, this proposal proposes a universal and low-cost passive indoor intrusion detection utilizing commodity WiFi devices based on channel state information. Firstly, we will extract the channel state information from the commodity WiFi wireless card and detect the target of different moving speeds in indoor scenario based on supervised learning methods. Secondly, we will explore the detection of the target of different moving speeds based on unsupervised learning methods. Finally, on the condition that there exists human motion we will extract gait information from channel state information and explore identity authentication based on gait information. The above works build the method of implementing indoor passive intrusion detection based on wireless channel state information, and promote its application and popularization in common indoor scenario.

无线被动室内入侵检测是在被动室内定位基础上，通过利用不同目标对无线信号传播的影响不同来实现入侵检测的技术。无线被动室内入侵检测由两个阶段组成：人体检测和身份认证。目前绝大多数人体检测和身份认证技术都需要很严格的假设条件，普适性不强，无法用于室内的入侵检测。针对该问题，本课题提出基于无线信道状态信息利用普通商用无线设备实现普适的低成本被动室内入侵检测。本课题首先从普通商用WiFi无线网卡中获取信道状态信息，采用有监督的方式检测室内不同移动速度的目标。在此基础上研究采用无监督的方式检测室内不同移动速度的目标。最后在检测到室内有人的情况下，从信道状态信息中提取出人走路的步态信息，研究基于步态指纹的身份认证。本课题将建立完善的基于无线信道状态信息的被动室内入侵检测技术，促进其在一般室内环境下应用的普及。

无线被动室内入侵检测是在被动室内定位基础上，通过利用不同目标对无线信号传播的影响不同来实现入侵检测的技术。无线被动室内入侵检测由两个阶段组成：人体检测和身份认证。目前绝大多数人体检测和身份认证技术都需要很严格的假设条件，普适性不强，无法用于室内的入侵检测。针对该问题，本课题提出基于无线信道状态信息利用普通商用无线设备实现普适的低成本被动室内入侵检测。本课题首先从普通商用WiFi无线网卡中获取信道状态信息，采用有监督的方式检测室内不同移动速度的目标。在此基础上研究采用无监督的方式检测室内不同移动速度的目标。最后在检测到室内有人的情况下，从信道状态信息中提取出人走路的步态信息，研究基于步态指纹的身份认证。本课题利用 CSI 的新型特征明显改善了被动室内入侵检测系统的健壮性，提高了系统在真实环境中的可用性和易用性，能够更好地应用于室内复杂环境中，具有重要的理论意义和实际应用价值。