# Joint Learning of Geometric and Probabilistic Constellation Shaping

@article{Stark2019JointLO,
title={Joint Learning of Geometric and Probabilistic Constellation Shaping},
author={Maximilian Stark and Fayccal Ait Aoudia and Jakob Hoydis},
journal={2019 IEEE Globecom Workshops (GC Wkshps)},
year={2019},
pages={1-6}
}
• Published 18 June 2019
• Computer Science
• 2019 IEEE Globecom Workshops (GC Wkshps)
The choice of constellations largely affects the performance of communication systems. When designing constellations, both the locations and probability of occurrence of the points can be optimized. These approaches are referred to as geometric and probabilistic shaping, respectively. Usually, the geometry of the constellation is fixed, e.g., quadrature amplitude modulation (QAM) is used. In such cases, the achievable information rate can still be improved by probabilistic shaping. In this work…
27 Citations

## Figures from this paper

Joint Learning of Probabilistic and Geometric Shaping for Coded Modulation Systems
• Computer Science
GLOBECOM 2020 - 2020 IEEE Global Communications Conference
• 2020
We introduce a trainable coded modulation scheme that enables joint optimization of the bit-wise mutual information (BMI) through probabilistic shaping, geometric shaping, bit labeling, and demapping
Constellation Shaping in Optical Communication Systems
This thesis investigates the performance of constellation shaping methods including geometric shaping (GS) and probabilistic shaping (PS) in coherent fiber-optic systems and finds that while PSQAM outperforms the uniform QAM in the linear regime, uniformQAM can achieve better performance at the optimum power in the presence of transmitter or channel nonlinearities.
Performance-Enhanced DMT System With Joint Precoding and Probabilistic Constellation Shaping
• Computer Science
IEEE Photonics Journal
• 2021
Joint precoding and PCS significantly improve the resistance to nonlinear distortions for DMT systems and adopt the probabilistic constellation shaping (PCS) technique to reduce the probability of constellation points with high power, resulting in the overall performance improvement.
Joint Constellation Design and Multiuser Detection for Grant-Free NOMA
• Computer Science
IEEE Transactions on Wireless Communications
• 2022
A DL-based multi-task variational autoencoder (Mul-VAE) that adopts a variational AUTO network to optimize the distribution of the constellation points and results reveal that the proposed method enables significant gains compared to state-of-the-art techniques.
Discriminative Mutual Information Estimation for the Design of Channel Capacity Driven Autoencoders
• Computer Science
ArXiv
• 2021
A set of novel discriminative mutual information estimators are presented and it is discussed how to exploit them to design capacity-approaching codes and ultimately estimate the channel capacity.
• Computer Science
2020 12th International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP)
• 2020
End-to-end learning is presented, a recent technique in communications to learn optimal transmitter and receiver architectures based on deep neural networks (DNNs) architectures, and cases in which this technique has been used to design constellations, where a mathematical analysis is repressed due to the channel model intractability.
Capacity-Approaching Autoencoders for Communications
• Computer Science
ArXiv
• 2020
This paper addresses the challenge of designing capacity-approaching codes by incorporating the presence of the communication channel into a novel loss function for the autoencoder training by exploiting the mutual information between the transmitted and received signals as a regularization term in the cross-entropy loss function.
Model-Based Deep Learning of Joint Probabilistic and Geometric Shaping for Optical Communication
• Computer Science, Physics
ArXiv
• 2022
The optimized constellation shaping outperforms the 256 QAM Maxwell-Boltzmann probabilistic distribution with extra 0.05 bits/4D-symbol mutual information for 64 GBd transmission over 170 km SMF link.
Capacity-Driven Autoencoders for Communications
• Computer Science
IEEE Open Journal of the Communications Society
• 2021
This paper addresses the challenge of designing capacity-approaching codes by incorporating the presence of the communication channel into a novel loss function for the autoencoder training by exploiting the mutual information between the transmitted and received signals as a regularization term in the cross-entropy loss function.
Trainable Communication Systems: Concepts and Prototype
• Computer Science
IEEE Transactions on Communications
• 2020
We consider a trainable point-to-point communication system, where both transmitter and receiver are implemented as neural networks (NNs), and demonstrate that training on the bit-wise mutual

## References

SHOWING 1-10 OF 20 REFERENCES
Geometric Constellation Shaping for Fiber Optic Communication Systems via End-to-end Learning
• Physics, Computer Science
ArXiv
• 2018
By embedding a differentiable fiber channel model within two neural networks, the learning algorithm is optimizing for a geometric constellation shape, and allows joint optimization of system parameters, such as the optimal launch power, simultaneously with the constellation shape.
On Probabilistic Shaping of Quadrature Amplitude Modulation for the Nonlinear Fiber Channel
• Physics
Journal of Lightwave Technology
• 2016
The results show that, for a fixed average optical launch power, a shaping gain is obtained for the noise contributions from fiber amplifiers and modulation-independent nonlinear interference (NLI), whereas shaping simultaneously causes a penalty as it leads to an increased NLI.
Optimal nonuniform signaling for Gaussian channels
• Computer Science
IEEE Trans. Inf. Theory
• 1993
Variable-rate data transmission schemes in which constellation points are selected according to a nonuniform probability distribution are studied and prefix codes can be designed that approach the optimal performance.
Deep Learning for Channel Coding via Neural Mutual Information Estimation
• Computer Science
2019 IEEE 20th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC)
• 2019
This work uses a recently proposed neural estimator of mutual information to optimize the encoder for a maximized mutual information, only relying on channel samples, and shows that this approach achieves the same performance as state-of-the-art end-to-end learning with perfect channel model knowledge.
Communication Algorithms via Deep Learning
• Computer Science
ICLR
• 2018
It is shown that creatively designed and trained RNN architectures can decode well known sequential codes such as the convolutional and turbo codes with close to optimal performance on the additive white Gaussian noise (AWGN) channel.
Bandwidth Efficient and Rate-Matched Low-Density Parity-Check Coded Modulation
• Computer Science
IEEE Transactions on Communications
• 2015
A new coded modulation scheme is proposed that operates within less than 1.1 dB of the AWGN capacity 1/2 log2(1 + SNR) at any spectral efficiency between 1 and 5 bits/s/Hz by using only 5 modes.
Physical layer deep learning of encodings for the MIMO fading channel
• Computer Science
2017 55th Annual Allerton Conference on Communication, Control, and Computing (Allerton)
• 2017
This work introduces a novel physical layer scheme for Multiple Input Multiple Output (MIMO) communications based on unsupervised deep learning using an autoencoder and discusses how the scheme can be easily adapted for open-loop and closed-loop operation in spatial multiplexing modes as well as spatial diversity modes.
Massive MIMO Networks: Spectral, Energy, and Hardware Efficiency
• Computer Science
Found. Trends Signal Process.
• 2017
This monograph summarizes many years of research insights in a clear and self-contained way and providest the reader with the necessary knowledge and mathematical toolsto carry out independent research in this area.
Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
• Computer Science
ArXiv
• 2013
This work considers a small-scale version of {\em conditional computation}, where sparse stochastic units form a distributed representation of gaters that can turn off in combinatorially many ways large chunks of the computation performed in the rest of the neural network.
A Novel PAPR Reduction Scheme for OFDM System Based on Deep Learning