Research On Key Technology Of Transparent Sensing Via Channel State Information

Transparent sensing is one of the research hotpots during the last few years.It is widely applied to various applications like intrude detection,smart home,activity analysis and health care.However,the current solutions suffer from several major demerits,e.g.,heavily relying on the environment,poor user experience or expensive hardware device,Therefore,their adaptability remains a major issue.Transparent sensing via using channel state information as media has more diversity and humanization in real-world scenarios.In this dissertation,we explore channel state information collected by Wi-Fi for building a series of key transparent sensing technologies.In particular,this dissertation uses a lightweight algorithm that requires only a small training in sensing of whole-body activities.To sense partial-body movements,the Gaussian Mixed Model(GMM)algorithm is used to extract the action without the need of training data in advance.Experiments show that our proposed algorithm is effective and robust.In summary,the main contributions of this article are as follows:Frist,we conduct an empirical study on channel state information for both whole-body and partial-body movements.Second,to deal with the complex and changeable channel state information,we propose a lightweight solution to select high quality subcarriers and reduce the amount of operations on data processing.Third,a background extracting algorithm based on the Gaussian Mixed Model is proposed to identify the movement of people during sleep,which can be used to detect the quality of sleep.