Institute of Telecommunications


Information on some of our current research activities.

Machine Learning for Communications

Can we "learn" to communicate?

Inspired by Shannon’s groundbreaking work on the fundamental limits of communications, we seek to solve the task of “reproducing at one point either exactly or approximately a message selected at another point”.  In other words, we seek to reliably transmit a message from a source to a destination over a channel using a transmitter and a receiver.

We propose a thinking-outside-the-box solution where we reinterpret transceiver signal-processing blocks (e.g., quantization, error correcting coding, modulation, detection) as neural networks, enabling data-driven communications systems that perpetually learn and adapt to any environment.

We built the world’s first over-the-air communications system that consists solely of neural networks. In this spirit, we propose one holistic block that does it all using joint signal-processing for detection, synchronization and decoding so that the information is completely transmitted within a single sequence (

Current Research

Bit-wise Autoencoder Communication System Training

This video shows the training process of a bit-wise autoencoder communications system. The system tries to find optimal signals to transmit k=4 bit within n=2 real valued channel uses at the trained SNR of 3.85dB. On the left side one can see the currently used constellation and on the right side one can see the mutual information obtained at receiver side. The bottom heat map plots show the current decision regions of the learned receiver, where pink regions indicate a strong decision towards bit 0 and blue regions indicate a strong decision towards bit 1. It can be seen, that the system finally converges towards a constellation that differs from classical modulation schemes, like the comparable 16-QAM constellation, and is thereby able to transmit more information at this operational point. While these gains are well known in communications research as so called shaping gains, the possible game changer of a neural network based autoencoder system is, that it is able to optimize its signals over every kind of channel, including all unknown insufficiencies using data based training strategies.

For more information see the paper "Trainable Communication Systems: Concepts and Prototype" by S. Cammerer, F. A. Aoudia, S. Dörner, M. Stark, J. Hoydis and S. ten Brink at where these results are based on.


Stephan ten Brink
Prof. Dr.-Ing.

Stephan ten Brink


Christian  Senger

Christian Senger

Deputy Director

To the top of the page