The measurement accuracy in non-contact vital
signs detection can be drastically affected by strong interferences,
such as multiple targets in continuous wave radar system and
random body movements (RBMs). In this paper, a 77GHz linear-
frequency-modulated continuous-wave (LFMCW) radar system
with multiple receiver channels is firstly investigated to migrate
the interferences from multiple targets, and then a multi-channel
Kalman smoother (MCKS) is proposed to alleviate the effect of
RBMs and acquire an enhanced outlier-robust phase history of
vital signs for the subsequent accurate estimates of the respiration
rate (RR) and heart rate (HR). Simulations demonstrate that the
local SNRs around the RR and HR of the results obtained by
MCKS are 9dB and 1dB higher than those obtained by multi-
channel average (MCA), respectively, and the detection error
rates are less than 10% for RR and less than 6% for HR in multi-
target detection with RBMs environment in the real experiments.