Research On Key Technologies Of Transparent Context-awareness Driven By Channel State Information

Pervasive computing based on "context-awareness" has been widely used in intelligent healthcare systems.At present,mainstream methods of context-aware computing can be roughly divided into two categories,i.e.sensor-based and vision-based.There still exist many problems such as privacy issues,specialized devices,limited sensing area and so on.To this end,we introduce the key technologies of transparent sensing via Channel State Information(CSI).We simplify the human-computer interaction by obtaining the context information of human physiological and psychological state in a contactless way.In summary,the main work of this dissertation is listed as follows:1.We design the Sleep Guardian to capture subtle human motions during sleeping.The key idea of the system is to build a Gaussian Mixture Model(GMM)based approach,which can adaptively model the still posture and in-place motion in CSI extracted from Wi Fi signal.We evaluate the performance of Sleep Guardian in two real scenarios,i.e.office room and conference room.The average detection rate of motions is up to 99.01%,and the mean absolute error of motions is only 0.81 s in Sleep Guardian.2.We design the Emo Sense,an emotion sensing system based on the CSI.This system explores the physical expression corresponding to CSI via data mining technology,and leverages the Fresnel Zone mathematics model to strengthen the physical expression of the fingerprint information on the channel state.At the same time,we collect 3360 behavior-emotion wireless signal cases,and leverage the k-NN algorithm to recover the emotional statement from the physical expression.Emo Sense can achieve the average recognition rate of 93.33% among four emotions,i.e.happiness,sadness,anger,and fear.