An automatic detection scheme for periodic signals based on spectrum analyzer measurements

Discerning noise from signal is often required for a correct interpretation of a spectrum. This paper presents a method to automatically select the significant spectral components in spectra, measured with a spectrum analyzer. No prior knowledge about the measured signals is required, and only commercially available equipment is used. The detection algorithm is based on the relationship between the resolution bandwidth (RBW) and the noise level of the spectrum analyzer. It is divided in three steps. First, the spectrum is measured with a spectrum analyzer. Second, the same signal is measured a second time with a different RBW. Finally, from the division of both measurements, one can conclude at which frequencies a signal is present. A statistical framework for the algorithm will be derived. The method will be demonstrated on simulations and measurements.