面向高动态环境的非绑定感知关键技术

Today the contactless sensing has become a research hotspot in academia and industry with the advantages of no carrying devices and no intrusiveness. However, due to diverse application scenes, different human locations and uncertain human body quantity in the high-dynamic environment, the ubiquitous application of contactless sensing faces three serious challenges: the difficult signal attenuation estimation, the difficult signal superposition prediction, and the difficult composite signal decomposition. Because previous work does not fully consider the high dynamics characteristic of the ubiquitous environment, the contactless sensing has three serious problems , such as failing multi-scene sensing, abnormal multi-location sensing, and wrong multi-human sensing. As a result, this project focuses on the research of contactless sensing key technique in the high-dynamic environment. The research includes multi-scene adaptive sensing based on attention factor, multi-position sensing based on Fresnel zone model, and multi-person sensing based on tensor-based decomposition theory. It is intended to achieve research goals such as effective multi-scene sensing, accurate multi-position sensing, and real multi-person sensing. This topic aims to promote the ubiquitous application of high-adaptive contactless sensing.

当前非绑定感知以零设备携带、零用户侵扰等优势成为学术界和工业界的研究热点。然而由于普适环境中存在场景多样性、位置差异性以及人数不确定性等高动态现象，导致非绑定感知面临信号衰减难估量、信号增消难预测以及信号分解难处理等严峻挑战。而当已有的工作未充分考虑普适环境的高动态性，导致非绑定感知存在多场景感知失效、多位置感知失常以及多人体感知失真等严重问题。因此，本课题针对高动态环境下非绑定感知关键技术展开深入科学研究，研究内容包括基于衰减因子的多场景自适应感知、基于菲涅尔区模型的多位置感知以及基于张量分解理论的多人体感知。本课题拟实现多场景有效感知、多位置精准感知、多人员准确感知等研究目标，促进高适应性的非绑定感知应用的普及。

当前非绑定感知以零设备携带、零用户侵扰等优势成为学术界和工业界的研究热点。由于普适环境中存在场景结构多样性、人体位置差异性以及人数不确定性等高动态现象，非绑定感知面临着信号衰减难估量、信号增消难预测以及信号分解难处理等严峻挑战。普适环境的高动态性导致非绑定感知出现多场景感知失效、多位置感知失常以及多人体感知失真等鲁棒性问题。因此，本课题针对高动态环境下非绑定感知关键技术展开深入科学研究，提出了基于近场人体位置感知、高鲁棒性自适应人体运动感知以及基于动态时间规整的体操动作规范性评估感知等多个模型，成功刻画人体状态与信号变化之间的关系、消除非绑定感知对现场勘测工作的依赖以及从低分辨率WiFi中提取多频信号特征。我们在多场景下开展了大量科学实验，测试结果显示本课题提出的感知模型能够取得较高的精度。在课题负责人和参与者的共同努力下，我们按期完成了研究计划中预定的研究目标，还在移动应用安全性分析、终端网络可靠性、WiFi热点安全性等相关移动生态研究中取得了一系列成果。我们累计发表学术论文9篇，并有5篇论文已被CCF A类会议/期刊发表，申请发明专利2项，多项研究成果被小米、腾讯等公司实际采用。