Digital modulation classification using constellation shape

Constellation diagram is a traditional and powerful tool for design and evaluation of digital modulations. In this work we propose to use constellation shape as a robust signature for digital modulation recognition. We represent the transmitted "information" by the geometry of the constellation. Received information is in turn the recovered constellation shape that is deformed by noise, channel and receiver implementation. We first demonstrate that fuzzy c-means clustering is capable of robust recovery of the unknown constellation. To perform Bayesian inference, the reconstructed constellation is modeled by a discrete multiple-valued nonhomogenous spatial random field. For candidate modulations, their corresponding random fields are modeled off-line. The unknown constellation shape is then classified by an ML rule based on the preceding model building phase. The algorithm is applicable to digital modulations of arbitrary size and dimensionality. (C) 2000 Published by Elsevier Science B.V. All rights reserved.