An 'Internet of Ears' for crowd-aware smart buildings based on sparse sensor networks
This paper introduces sparse low-power sensor networks using passive seismic sensors for crowd-aware smart buildings, resulting in an 'Internet of Ears (IoE)' that can hierarchically detect human footsteps, estimate walk direction, and track person position. An energy-based thresholding footstep detector was developed using wavelet transform for pre-filtering the noise-prone ambient vibrations detected by the sensors. An efficient time delay estimate based on maximum power criterion is proposed for beamforming. Indoor occupant localization based on a least squares (LS) method and using improved time delay estimation is also briefly introduced. Experimental results in two different buildings show the effectiveness of the footstep detection and occupant localization methods, with detection accuracy of 98.3% and average localization error <25 cm.
