Multiple signal classification for self-mixing flowmetry.

ABSTRACT

For the first time to our knowledge, we apply the multiple signal classification (MUSIC) algorithm to signals obtained from a self-mixing flow sensor. We find that MUSIC accurately extracts the fluid velocity and exhibits a markedly better signal-to-noise ratio (SNR) than the commonly used fast Fourier transform (FFT) method. We compare the performance of the MUSIC and FFT methods for three decades of scatterer concentration and fluid velocities from 0.5 to 50 mm/s. MUSIC provided better linearity than the FFT and was able to accurately function over a wider range of algorithm parameters. MUSIC exhibited excellent linearity and SNR even at low scatterer concentration, at which the FFT's SNR decreased to impractical levels. This makes MUSIC a particularly attractive method for flow measurement systems with a low density of scatterers such as microfluidic and nanofluidic systems and blood flow in capillaries.