CSI Based Human Tracking System In Indoor Environment

Human tracking in indoor environment plays an important role in intelligent housing system that enables the smart appliances to provide fine-grained services with the information of users' walking trajectories.This paper presents a device-free human tracking system only using existing commercial WiFi supported devices and smart appliances.We consider that a human walking trajectory consists of a series of moving behaviors and we can determine the trajectory by detect those behaviors.There are three types of moving behaviors defined in our works,Pass-Device,Pass-Room and Enter/Exit-Room who can be sensed according to their influence on the signals of neighbouring WiFi devices(we call these devices Check Points).A machine learning method,Support Vector Machine(SVM)is used to train a classifier to identify the above moving behaviors.Furthermore,an uniform trace recording format consisting of the related Check Points,types and occurrence time of the moving behaviors series,is designed for the comparison of two different walking trajectories.By calculating the similarity between the measured human trace and stored trace profiles in Trace Pool,we can determine the specific trajectory.Our experimental evaluation in three different scenarios shows that our approach can achieve over 95% average accuracy on moving behaviors identification and over 90% average accuracy on trace determination.