Performance Study of Algorithms for Detecting Pulsed Sinusoidal Interference in Microwave Radiometry

The performance of four algorithms for detecting the presence of pulsed sinusoidal interference in microwave radiometry is compared. The pulsed sinusoidal interference sources considered have unknown frequency, initial phase, amplitude, arrival time, and duration. The statistical properties of three of the algorithms are determined analytically, although numerical integrations are required in some cases in order to compute the obtained probabilities of detection. The performance of the fourth algorithm is evaluated using Monte Carlo procedures. Results show that three of the algorithms have a performance that is roughly comparable for the cases considered, while the fourth yields reduced sensitivity. A more detailed study, of two of the algorithms, a simple energy, detector called the pulse detection algorithm and a "peak" picking detector based on the use of data spectrograms, is also provided.