immortal3/AutoEncoder-Based-Communication-System

Models communications systems as autoencoders where transmitter and receiver are jointly optimized end-to-end using TensorFlow and Keras. Supports configurable (n,k) encoder architectures with dynamic training via Jupyter notebooks, generating bit-error-rate performance metrics and learned constellation diagrams across varying signal-to-noise ratios. Demonstrates how neural networks can learn optimal modulation and channel coding strategies directly from data rather than relying on traditional hand-crafted communication schemes.

135 stars. No commits in the last 6 months.