Performance Analysis of the Maximum Likelihood Estimation of Signal Period Length and Its Application in Heart Rate Estimation with Reduced Respiratory Influence

Abstract

The remote and non-contact monitoring of human respiration and heartbeat based on radars is a safe and convenient practice. However, how to accurately estimate the heart rate is still an open issue, because the heartbeat information in radar signals is affected by respiratory harmonics. In this paper, a maximum likelihood estimation was introduced to extract the heart rate from high-pass-filtered radar heartbeat waveforms where the low-frequency respiratory and heartbeat components were attenuated. The closed-form asymptotic estimation variance of the maximum likelihood estimator was derived to describe its performance in white Gaussian noise with a high signal-to-noise ratio (SNR). The proposed method was verified using two publicly available datasets and demonstrated superior performance compared to other methods. The estimation method and the asymptotic estimation variance here described are also applicable for signal period estimation in other settings with similar conditions.