Automatic Modulation Classification for Cognitive Radio

A cognitive radio has to know which signals are already on the air before it may transmit. This project teaches a computer to recognise the modulation of a received radio signal from 128 complex baseband (I/Q) samples: eight digital schemes (BPSK, QPSK, 8PSK, PAM4, QAM16, QAM64, CPFSK, GFSK) and three analog ones (WBFM, AM-DSB, AM-SSB), at signal-to-noise ratios from -20 to +18 dB. Classical higher-order-cumulant features with SVM and gradient-boosting classifiers are compared with a 2-D convolutional neural network and an LSTM on the raw samples, using the real RadioML 2016.10A dataset (220,000 frames). A web application classifies uploaded GNU Radio captures, GNU Octave re-checks the features, and a GNU Radio flowgraph and a complete MATLAB version are included.

Category: Electronics and Communication Engineering (Wireless Communication, Cognitive Radio, Signal Processing, Deep Learning, GNU Radio, GNU Octave, MATLAB, Python)

Electronics & Communication Engineering Project E01 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.