Detection of a low level signal in noise by tegulometric methods.

Abstract

A novel and sensitive method is described for the analysis and detection of weak signals generated by many sources and highly contaminated by noise. This process yields a spectral display whose elements are related to the probability of occurrence of a frequency rather than to the power spectral density of conventional schemes. Thus, a weak signal and stronger signals may show equally if they appear over the same percentage of the analysis period. This process, called tegulometric analysis, is tested and improved in sensitivity for detection of very weak signals hidden in gaussian noise. The investigation shows that this method can be optimized to produce low level signal detections based upon short observational periods. The frequency resolution capability of the method is shown to be marked.