Classification of interference signals using advanced baseband statistics in pi/4-DQPSK systems

This paper presents a novel technique to classify an interfering signal that can be dual-path propagation, continuous-wave or noise. To identify the dominant interfering signal, a couple of unique statistical features have to be found in the baseband-signal. These features can be used to train a machine-learning algorithm to classify the interference signal. The presented algorithm is tested with a pi/4-DQPSK modulated continuous Terrestrial Trunked Radio (TETRA) downlink. The classification results of several machine learning algorithms are compared to find the most accurate solution.