Spectral estimation of terahertz time-domain signals using a simple parametric method

Abstract

A simple method for spectral estimation of terahertz signals using an autoregressive model is developed. The technique shows a better spectral variance and resolution over the use of the Fast Fourier Transform, which is commonly utilised in this kind of signal. The results are validated comparing the water vapour absorption lines and the standard deviation of a set of 24 measurements between methods. As a consequence of improving the signal variance, a better signal-to-noise ratio is achieved without enhancing the experiment. Such result also produces an increase in the signal bandwidth and less error when optical parameters are estimated.