Novel method for blind constellation detection using template based classifier for quadrature digital modulation schemes.

Abstract

Template based classifiers are very popular classifiers in biomedical and computer vision area. In computer vision they are typically used to understand or classify the scenes captured by camera. In biomedical field they are typically used to provide non-intrusive disease identification. For example to classify a tissue samples as cancerous or not. In this paper we apply the learnings of this area to device an algorithm for automatic modulation detection problem. Automatic modulation detection problem in communication receivers involves auto detecting the modulation scheme from the received samples at the communication receiver without the prior knowledge of the encoded modulation scheme. The method can efficiently recognize almost all quadrature digital modulation schemes and the accuracy rate is over 95% at the SNRs of 4.5 dB with as low as 512 bits. The performance of this algorithm is evaluated using simulations on LabVIEW.