信道状态信息的生命体征检测算法

Vital Signs Detection Algorithm Based on Channel State Information

针对传统的接触式呼吸检测设备易脱落、应用场景有限等问题,提出一种基于信道状态信息的生命体征检测算法。该算法以菲涅尔区模型为理论指导,通过商用无线保真(Wi-Fi)设备采集呼吸信道状态信息。首先,利用Hampel滤波器滤除信号衰落和多径效应引起的高频噪声;然后,以改进的阈值函数选取方法即开方法解决软硬阈值偏差恒定、不连续等问题,进一步滤除环境噪声;最后,通过去除直流分量、子载波选择和交叉平移点算法获取呼吸频率。此外,通过机器学习、深度学习算法实现了正常、异常和暂停3种不同呼吸状态的分类,全面反映了被检测人员的呼吸运动。实验结果表明:基于信道状态信息的生命体征检测算法准确率达到了92.5%,能够作为人体生命体征日常检测系统为用户提供健康参考。

Aiming at the problems that traditional contact breathing detection equipment was easy to fall off and has limited application scenarios,this paper proposed a vital sign detection algorithm based on channel state information.The algorithm was guided by the Fresnel zone model.The respiratory channel state information was collected by commercial wireless fidelity(Wi-Fi) equipment.Firstly,the Hampel filter was used to filter out the high-frequency noise caused by signal fading and multipath effects. Then, an improved threshold function selection method,that is open method,was used to solve the problems of constant and discontinuous soft and hard threshold deviations,and further filter out environmental noise.Finally,the respiratory frequency was obtained by removing the DC component,sub-carrier selection and cross-translation point algorithm. In addition,the classification of three different breathing states,normal,abnormal and paused,was realized through machine learning and deep learning algorithms,which fully reflected the breathing movement of the detected person.The experimental results show that the accuracy of the vital signs detection algorithm based on the channel state information reaches 92.5%,which can be used as a daily detection system for human vital signs to provide users with health reference.