beneboeck/wireless-chan-mod4ml — reverse-engineered prompt

Build me a research code project for wireless channel modeling for machine learning. I want to be able to generate wireless channel datasets in MATLAB, convert them for Python, then run simple learning and signal processing experiments on them.

Include MATLAB scripts for TDL and CDL channel data using the 5G Toolbox, and scenario data using QuaDRiGa for rural and urban setups. Make the model type, number of samples, and subcarriers easy to change. Save the data, then provide Python converters from mat files to npy files, plus a few small toy datasets so the project can run quickly.

Add Python command line scripts for CSI compression with an autoencoder, compression with PCA, channel estimation with LMMSE, and sample covariance Gaussian generation. Each script should take dataset name, train and test sizes, latent size or SNR where needed, and save results cleanly.

Please include a clear README with setup steps, example commands, dataset path configuration, citation text, and notes about external tools like QuaDRiGa and DeepMIMO.