Frequency Estimation of Human Presence Detection Signals From a Scanning-Beam Millimeter-Wave Correlation Radiometer

Abstract

The scanning-beam millimeter-wave correlation radiometer has recently been applied to human presence detection and classification. The natural frequency of the scanning-beam correlation radiometer signal, called the fringe frequency, results from human presence in the scanning beam and must be known in order to filter the signal and provide the best possible radiometric sensitivity. This letter presents new results in estimating the fringe frequency of a 27.4-GHz scanning-beam correlation radiometer by comparing different spectral estimation techniques, including CLEAN, RELAX, PHD, MUSIC, and ESPRIT. Experiments were performed with a human placed beyond the mean detection range of the sensor, resulting in signal power levels below the average power expected in an operational environment. Results show good performance for the CLEAN and RELAX algorithms at fringe frequencies in the range of 15-60 Hz and for the MUSIC and ESPRIT algorithms at fringe frequencies in the range of 60-155 Hz.