Decision fusion methods for automatic modulation classification with multiple sensors in multipath fading channels

Abstract

Cognitive radio networks (CRN) and wireless sensor networks (WSN) inherently deploy multiple sensors, thus providing an excellent basis for distributed signal sensing and modulation classification. Deploying cooperative multiple sensors offers considerable improvement of automatic modulation classification (AMC) performances, compared to single sensor deployment, especially in the presence of multipath fading. In this paper, different decision fusion methods for AMC with multiple sensors in multipath fading environment were presented and analyzed. Soft decision fusion and joint decision fusion were defined for AMC algorithm using fourth-order cumulants, and comparison with appropriate hard decision fusion methods for same AMC algorithm is performed. Classification performances of derived AMC schemes with fusion of decision, in terms of average probability of correct classification, are investigated and evaluated through Monte-Carlo simulations.