End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification

Roya Khanzadeh; Stefan Angerbauer; Andreas Springer; Werner Haselmayr

doi:10.1109/IEEECONF59524.2023.10476924

End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification

Department of Communications Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

Abstract

In this paper, we introduce a novel end-to-end deep learning procedure for communication systems, which is data- efficient and capable of dealing with infinite memory length of communication channels. Therefore, as opposed to recent works, we utilize a low-complexity algorithm to identify the communication channel. The channel model is obtained purely from data and its output is differentiable with respect to its input, which is a basic requirement for gradient-based optimization of the auto-encoder neural network. We study the performance of the algorithm for a variety of challenging channels from different domains of communication engineering showing the broad applicability of the proposed approach.

Original language	English
Title of host publication	Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023
Editors	Michael B. Matthews
Pages	40-46
Number of pages	7
ISBN (Electronic)	9798350325744
DOIs	https://doi.org/10.1109/IEEECONF59524.2023.10476924
Publication status	Published - Oct 2023

Publication series

Name	Conference Record - Asilomar Conference on Signals, Systems and Computers
ISSN (Print)	1058-6393

Fields of science

202038 Telecommunications
202 Electrical Engineering, Electronics, Information Engineering
202037 Signal processing

JKU Focus areas

Digital Transformation

Access to Document

10.1109/IEEECONF59524.2023.10476924

Cite this

Khanzadeh, R., Angerbauer, S., Springer, A., & Haselmayr, W. (2023). End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification. In M. B. Matthews (Ed.), Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023 (pp. 40-46). (Conference Record - Asilomar Conference on Signals, Systems and Computers). https://doi.org/10.1109/IEEECONF59524.2023.10476924

Khanzadeh, Roya ; Angerbauer, Stefan ; Springer, Andreas et al. / End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification. Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023. editor / Michael B. Matthews. 2023. pp. 40-46 (Conference Record - Asilomar Conference on Signals, Systems and Computers).

@inproceedings{aba911b15bf846e38066ecd72482695d,

title = "End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification",

abstract = "In this paper, we introduce a novel end-to-end deep learning procedure for communication systems, which is data- efficient and capable of dealing with infinite memory length of communication channels. Therefore, as opposed to recent works, we utilize a low-complexity algorithm to identify the communication channel. The channel model is obtained purely from data and its output is differentiable with respect to its input, which is a basic requirement for gradient-based optimization of the auto-encoder neural network. We study the performance of the algorithm for a variety of challenging channels from different domains of communication engineering showing the broad applicability of the proposed approach.",

author = "Roya Khanzadeh and Stefan Angerbauer and Andreas Springer and Werner Haselmayr",

year = "2023",

month = oct,

doi = "10.1109/IEEECONF59524.2023.10476924",

language = "English",

series = "Conference Record - Asilomar Conference on Signals, Systems and Computers",

pages = "40--46",

editor = "Matthews, \{Michael B.\}",

booktitle = "Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023",

}

Khanzadeh, R , Angerbauer, S , Springer, A & Haselmayr, W 2023, End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification. in MB Matthews (ed.), Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023. Conference Record - Asilomar Conference on Signals, Systems and Computers, pp. 40-46. https://doi.org/10.1109/IEEECONF59524.2023.10476924

End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification. / Khanzadeh, Roya ; Angerbauer, Stefan ; Springer, Andreas et al.
Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023. ed. / Michael B. Matthews. 2023. p. 40-46 (Conference Record - Asilomar Conference on Signals, Systems and Computers).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

TY - GEN

T1 - End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification

AU - Khanzadeh, Roya

AU - Angerbauer, Stefan

AU - Springer, Andreas

AU - Haselmayr, Werner

PY - 2023/10

Y1 - 2023/10

N2 - In this paper, we introduce a novel end-to-end deep learning procedure for communication systems, which is data- efficient and capable of dealing with infinite memory length of communication channels. Therefore, as opposed to recent works, we utilize a low-complexity algorithm to identify the communication channel. The channel model is obtained purely from data and its output is differentiable with respect to its input, which is a basic requirement for gradient-based optimization of the auto-encoder neural network. We study the performance of the algorithm for a variety of challenging channels from different domains of communication engineering showing the broad applicability of the proposed approach.

AB - In this paper, we introduce a novel end-to-end deep learning procedure for communication systems, which is data- efficient and capable of dealing with infinite memory length of communication channels. Therefore, as opposed to recent works, we utilize a low-complexity algorithm to identify the communication channel. The channel model is obtained purely from data and its output is differentiable with respect to its input, which is a basic requirement for gradient-based optimization of the auto-encoder neural network. We study the performance of the algorithm for a variety of challenging channels from different domains of communication engineering showing the broad applicability of the proposed approach.

UR - https://www.scopus.com/pages/publications/85190362569

U2 - 10.1109/IEEECONF59524.2023.10476924

DO - 10.1109/IEEECONF59524.2023.10476924

M3 - Conference proceedings

T3 - Conference Record - Asilomar Conference on Signals, Systems and Computers

SP - 40

EP - 46

BT - Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023

A2 - Matthews, Michael B.

ER -

Khanzadeh R , Angerbauer S , Springer A , Haselmayr W. End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification. In Matthews MB, editor, Conference Record of the 57th Asilomar Conference on Signals, Systems and Computers, ACSSC 2023. 2023. p. 40-46. (Conference Record - Asilomar Conference on Signals, Systems and Computers). doi: 10.1109/IEEECONF59524.2023.10476924

End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification

Abstract

Publication series

Fields of science

JKU Focus areas

Access to Document

Other files and links

JKU LIT SAL eSPML Lab

Cite this

End-to-End Learning of Communication Systems with Novel Data-Efficient IIR Channel Identification

Abstract

Publication series

Fields of science

JKU Focus areas

Access to Document

Other files and links

Projects

JKU LIT SAL eSPML Lab

Cite this