Synthetic Alpha-Beta Brainwave Framework for BCI Signal Classification

This study introduces a compact and reproducible framework for simulating multichannel electroencephalogram (EEG) signals in the alpha (8\u201312\u202fHz) and beta (13\u201330\u202fHz) bands, augmented with realistic noise and blink-like artifacts. A lightweight one-dimensional convolutional neural network (1D-CNN) is trained on 200 synthetic samples (8 channels, 30 seconds each) and achieves 100% test accuracy in distinguishing alpha from beta patterns. Comparative analysis with classical baselines\u2014including power spectral density (PSD) features with support vector machines (SVM) and shallow multilayer perceptrons (MLP)\u2014demonstrates superior efficiency (~10K parameters, ~5\u202fms inference latency) and robustness. The end-to-end pipeline, from signal synthesis to model evaluation, is open-source and optimized for rapid brain\u2013computer interface (BCI) prototyping under controlled conditions.