Research On Data Gathering And Recognition With Wireless Signals

With the development of sensors and wireless communication technologies,wireless networks have penetrated into many aspects of people's lives,such as environmental detection,target location,activity recognition and so on.However,existing wireless networks are also affected and constrained by energy and indoor propagation environments during deployment and application.This thesis focuses on compressive data gathering and human activity recognition in wireless network,which aims to save energy consumption and the activity recognition by wireless signals through the wall.Specifically,the existing compression-based wireless data gathering method mainly considered the reliable network environment.However,packet loss is a common phenomenon in wireless networks.Meanwhile,the wireless signal is easily affected by the indoor environment and obstacles during indoor propagation,the indoor propagation model cannot be accurately characterized.To address the above challenges,this thesis focuses on extending the network running time and improving the indoor wireless signal recognition of human activities.The main contents are as follows:(1)For wireless network data collection under unreliable links,we proposed a sparse random data gathering approach with packet loss consideration,which can effectively cope with the impact of network packet loss on the recovery quality of sensory data in wireless sensor network data gathering.Meanwhile,to satisfy the conditional requirements of compressed sensing theory for sparse projection matrix and sparse representation basis,we design a low-coherence sparse representation basis with sparse projection matrix constraint.The experimental results of real network data show that the proposed joint sparse projection data gathering approach can recover and reconstruct the sensory data accurately.In addition,the simulation results also show that the proposed scheme can effectively reduce network energy consumption compared with the existing compressed sensing data gathering approaches in wireless sensor network data gathering.(2)For Wi-Fi based indoor human activity recognition,we proposed TW-See,a device-free passive human activity recognition system with Wi-Fi signals,which does not require any dedicated device and meets the scenarios of the signals through the wall.In order to implement the TW-See system,we propose the Or-PCA technology to obtain the correlation between the human activities and it resulting changes in channel state information values.Meanwhile,this thesis also proposes a normalized variance sliding window algorithm,which uses it to identify the true start and end points of the activities and segment the waveform where the activities is located.Our experimental results show that TW-See system achieves an average accuracy of 94.46% when the signals pass through the concrete wall.