Singular Spectrum Analysis-Based Algorithm for Vitality Monitoring Using M-Sequence UWB Sensor

Recently, noninvasive vital signs detection is widely used in through the wall (TTW) human detection and healthcare monitoring. In indoors and TTW environments, low signal to noise ratio make the detection of the weak vital signs difficult, and causes errors in the estimation of heartbeat, respiration rate. In this paper, different techniques are applied to the signal collected using a M-sequence ultra wideband sensor from the human subject, in order to suppress the clutter, detect the target range and estimate the heartbeat and respiration rate. First, moving average method is used to remove the stationary clutter. Next, Singular Value Decomposition (SVD) is applied to cancel the non-stationary clutter. Then, variance analysis is applied to estimate the range; when range has been detected, one dimensional Singular Spectrum Analysis (SSA) is applied to the range bin containing the subject data. It is observed that SSA completely succeeds to decompose the signal into heartbeat and respiration signal and a collection of unwanted signals as clutter, noise, and harmonics. Finally, Fast Fourier transform (FFT) is used for the estimation of the heartbeat and respiration rate. Comparison results show that the proposed algorithm outperforms a reference SVD based algorithms for heartbeat and respiration rate estimation.