基于Wi-Fi CSI的多人行为鲁棒感知模型与方法研究

Wi-Fi无线感知具有普适程度高、感知范围广、感知成本低、不侵扰用户、不泄露隐私等特点，是实现人类行为感知的理想形式，成为智能感知、普适计算等领域的研究热点。现有研究主要面向单人场景，据此建立的理论与方法体系无法有效支持多人行为感知和识别。本项目针对由于Wi-Fi无线感知系统有限能力和多人行为感知复杂需求之间“供需失衡”导致的研究挑战，围绕多人行为感知的鲁棒性问题，重点研究基于Wi-Fi多参数/多载波/多设备协同互补的多人行为鲁棒感知模型和方法。具体研究内容包括：1）结合Wi-Fi信号多径叠加与多载波传输特点，构建时频混叠复合信号可分离性判定模型；2）基于不同信道参数的差异性，研究复合回波信号实时有效分离方法；3）利用多视图感知数据的互补性，研究场景自适应的多人行为鲁棒识别方法。本研究意义在于探索并建立适应多人场景的无线感知理论和方法，为公共安全、灾难应急、智慧健康等重要应用需求提供支撑。

Wi-Fi based wireless sensing has the characteristics of high pervasiveness, wide sensing range, low sensing cost, non-intrusive to users, and no leakage of privacy, which is an ideal way to realize human behavior recognition and has become a hot topic in the research fields of intelligent sensing and pervasive computing. Existing studies mainly focus on single-person scenarios, and the established theories and methods cannot effectively support multi-person behavior sensing and recognition. To address the challenge due to the "supply-demand imbalance" between the limited capabilities of the Wi-Fi based wireless sensing system and the complex needs of multi-person behavior recognition, this project will mainly focus on exploring the cooperative and complementary characteristics of multiple parameters / multiple subcarriers / multiple links, aiming at developing a robust multi-person behavior sensing and recognition framework. In particular, we will study: 1) the separability evaluation model for time-frequency aliased composite signals based on the characteristics of multi-path superposition and multi-carrier transmission of Wi-Fi signals, 2) the separation algorithm for composite echo signals based on the cooperativeness of different Wi-Fi channel parameters, and 3) the recognition algorithm for multi-person behaviors leveraging the complementarity of multi-view sensing data. The significance of this project lies in the development of a wireless sensing framework that is suitable for robust recognition of human behaviors in multi-person scenarios, and thus providing effective support for applications such as public security, disaster relief, and smart healthcare.