Sampling and algorithms aid modulation recognition

Techniques of modulation recognition in electronic surveillance systems are described. Emphasis is given to a novel approach employing sampling techniques and a simple algorithm to enhance recognition confidence when the SNR is high. A block diagram of a modulation recognizer system incorporating the new design approach is given to illustrate the five basic functions of the system, including: recovery of the envelope and zero-crossings of a received signal; collection of envelope and zero crossing samples; and computation. The recognition algorithm built into the system is based on the 'concept of similarity' from pattern recognition theory. The procedures employed to classify an unknown signal are also described, and some factors affecting recognizer performance are considered. Among the factors discussed are: excessive noise in the CW AM, FM, or OOK bands; and nonlinearities in the receiver front end and multipath. A group of probability matrices characterizing the recognizer performance is provided.