An Introduction to Deep Learning for the Physical Layer (2017-02-02T00:00:00.000000Z)

TL;DR

A fundamental new way to think about communications system design as an end-to-end reconstruction task that seeks to jointly optimize transmitter and receiver components in a single process is developed.

Abstract

We present and discuss several novel applications of deep learning for the physical layer. By interpreting a communications system as an autoencoder, we develop a fundamental new way to think about communications system design as an end-to-end reconstruction task that seeks to jointly optimize transmitter and receiver components in a single process. We show how this idea can be extended to networks of multiple transmitters and receivers and present the concept of radio transformer networks as a means to incorporate expert domain knowledge in the machine learning model. Lastly, we demonstrate the application of convolutional neural networks on raw IQ samples for modulation classification which achieves competitive accuracy with respect to traditional schemes relying on expert features. This paper is concluded with a discussion of open challenges and areas for future investigation.

Authors

Tim O'Shea

3 papers

J. Hoydis

3 papers

TL;DR

Abstract

Authors

References77 items

Learning to optimize: Training deep neural networks for wireless resource management

Deep MIMO detection

Learning to Optimize: Training Deep Neural Networks for Interference Management

Detection Algorithms for Communication Systems Using Deep Learning

RNN Decoding of Linear Block Codes

Scaling Deep Learning-Based Decoding of Polar Codes via Partitioning

On deep learning-based channel decoding

Learning to invert: Signal recovery via Deep Convolutional Networks

Deep Neural Networks to Enable Real-time Multimessenger Astrophysics

Automatic modulation classification based on high order cumulants and hierarchical polynomial classifiers

Learning Radio Resource Management in 5G Networks: Framework, Opportunities and Challenges

Reinforcement Learning with Unsupervised Auxiliary Tasks

Supervised-Learning-Aided Communication Framework for MIMO Systems With Low-Resolution ADCs

Blind detection for MIMO systems with low-resolution ADCs using supervised learning

Learning to Protect Communications with Adversarial Neural Cryptography

Hybrid computing using a neural network with dynamic external memory

Emotion recognition using wireless signals

Learning to communicate: Channel auto-encoders, domain specific regularizers, and attention

Onsager-corrected deep learning for sparse linear inverse problems

Learning to decode linear codes using deep learning

Theano: A Python framework for fast computation of mathematical expressions

Unsupervised representation learning of structured radio communication signals

Convolutional Radio Modulation Recognition Networks

Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural Networks

MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems

Neural Programmer-Interpreters

Capturing the human figure through a wall

Cognition-Based Networks: A New Perspective on Network Optimization Using Learning and Distributed Intelligence

Spatial Transformer Networks

Reinforcement Learning Neural Turing Machines - Revised

Human-level control through deep reinforcement learning

Gradient-based Hyperparameter Optimization through Reversible Learning

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

Adam: A Method for Stochastic Optimization

Deep Unfolding: Model-Based Inspiration of Novel Deep Architectures

Very Deep Convolutional Networks for Large-Scale Image Recognition

Caffe: Convolutional Architecture for Fast Feature Embedding

Identifying and attacking the saddle point problem in high-dimensional non-convex optimization

A Survey on Machine-Learning Techniques in Cognitive Radios

Large Scale Distributed Deep Networks

Random Search for Hyper-Parameter Optimization

Wirtinger Calculus Based Gradient Descent and Levenberg-Marquardt Learning Algorithms in Complex-Valued Neural Networks

Scikit-learn: Machine Learning in Python

A Survey on Transfer Learning

Rectified Linear Units Improve Restricted Boltzmann Machines

Large-scale deep unsupervised learning using graphics processors

RF Imperfections in High-rate Wireless Systems

A Fast Learning Algorithm for Deep Belief Nets

A new approach to signal classification using spectral correlation and neural networks

Applications of neural networks to digital communications - a survey

Object recognition from local scale-invariant features

Algorithms for automatic modulation recognition of communication signals

Digital communication receivers - synchronization, channel estimation, and signal processing

Wireless Communications: Principles and Practice

Joint source/channel coding for wireless channels

On the computational power of neural nets

8-PSK trellis codes for a Rayleigh channel

Multilayer feedforward networks are universal approximators

Dynamic system identification experiment design and data analysis

A mathematical theory of communication

GENERATIVE ADVERSARIAL NETS

Deep Learning

“Artiﬁcial neural networks for linear and non-linear impairment mitigation in high-baudrate IM/DD systems,”

IEEE Access Special Section Editorial: Artificial Intelligence Enabled Networking

keras

Dropout: a simple way to prevent neural networks from overfitting

Wireless Communications: Principles and Practice (2nd Edition) by

Torch7: A Matlab-like Environment for Machine Learning

Improving the speed of neural networks on CPUs

Visualizing Data using t-SNE

Foreword in "RF imperfections in high-rate wireless systems: impact and digital compensation"

Iterative Receiver Design

Complex-Valued Neural Networks

Optical Communications, 2nd ed

Generalization and network design strategies

Source Code