Joint modulation format/bit-rate classification and signal-to-noise ratio estimation in multipath fading channels using deep machine learning

Abstract

A novel algorithm for simultaneous modulation format/bit-rate classification and non-data-aided (NDA) signal-to-noise ratio (SNR) estimation in multipath fading channels by applying deep machine learning-based pattern recognition on signals’ asynchronous delay-tap plots (ADTPs) is proposed. The results for three widely-used modulation formats at two different bit-rates demonstrate classification accuracy of 99.8%. In addition, NDA SNR estimation over a wide range of 0−30 dB is shown with mean error of 1 dB. The proposed method requires low-speed, asynchronous sampling of signal and is thus ideal for low-cost multiparameter estimation under real-world channel conditions.