Modulation classification for burst-mode QAM signals in multipath fading channels

– We present a new and efficient method for identifying the modulation type of a bursty QAM signal in the presence of additive white Gaussian noise (AWGN) in unknown fading channels and with unknown carrier phase. Our approach is based on an iterative combination of blind equalization and soft-clustering techniques, utilizes the constant modulus algorithm (CMA) for an initial reconstruction of the received signal constellation, and employs a nonparametric soft clustering method to identify a small set of possible modulations based on the partition coefficient (PC) criterion. We then use an iterative approach in utilizing the decision adjusted modulus algorithm (DAMA) to refine our choices until we reach two hypotheses. Besides, we propose a parallel implementation of the proposed method, and provide an asymptotic analysis as well as simulation results to demonstrate the validity and usefulness of the proposed method. Simulation results indicate that our proposed method achieves a significant gain as compared to the cumulant-based approach for burst mode transmissions in fading channels.