Options
On neural network data predistortion and constellation shaping for satellite channels
Citation Link: https://doi.org/10.15480/882.15940
Publikationstyp
Conference Poster not in Proceedings
Date Issued
2025
Sprache
English
TORE-DOI
Citation
14th International ITG Conference on Systems, Communications and Coding, SCC 2025
Contribution to Conference
Peer Reviewed
true
In communication over transparent satellite payloads,
traveling wave-tube amplifier (TWTA) cause nonlinear distortion
to the transmit signal. Together with filters involved in the
transmission chain, the TWTA causes distorted decision regions
(warping) and intersymbol interference (ISI) at the matched
filter output. This nonlinear channel can be compensated by
means of data predistortion. The remaining interference results
can be interpreted as a channel and the transmit constellation
can be optimized for the joint channel with influences from
additive white Gaussian noise (AWGN) and phase noise. This
work shows how constellation shaping and data predistortion
can be designed using neural networks. Particularly, a novel data
predistortion algorithm based on a recurrent neural network
(RNN) which accounts for the ISI is presented. For constellation
shaping, different constrained optimization techniques are
presented, including quadrant-symmetric optimization, spiral
constellations and amplitude and phase-shift keying (APSK)
constellations. All constellations are shaped by implementing the
communication chain as an autoencoder, which is trained on
a channel with the abovementioned impairments, which occur
in satellite links. Resulting constellations are presented together
with data predistortion, bit error rate (BER) and information
rates.
traveling wave-tube amplifier (TWTA) cause nonlinear distortion
to the transmit signal. Together with filters involved in the
transmission chain, the TWTA causes distorted decision regions
(warping) and intersymbol interference (ISI) at the matched
filter output. This nonlinear channel can be compensated by
means of data predistortion. The remaining interference results
can be interpreted as a channel and the transmit constellation
can be optimized for the joint channel with influences from
additive white Gaussian noise (AWGN) and phase noise. This
work shows how constellation shaping and data predistortion
can be designed using neural networks. Particularly, a novel data
predistortion algorithm based on a recurrent neural network
(RNN) which accounts for the ISI is presented. For constellation
shaping, different constrained optimization techniques are
presented, including quadrant-symmetric optimization, spiral
constellations and amplitude and phase-shift keying (APSK)
constellations. All constellations are shaped by implementing the
communication chain as an autoencoder, which is trained on
a channel with the abovementioned impairments, which occur
in satellite links. Resulting constellations are presented together
with data predistortion, bit error rate (BER) and information
rates.
Subjects
digital predistortion
satellite communications
constellation shaping
autoencoders
DDC Class
621.3: Electrical Engineering, Electronic Engineering
Publisher‘s Creditline
Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting / republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to server or lists, or reuse of any copyrighted component of this work in other works
Publication version
acceptedVersion
Loading...
Name
Kopyto_Bauch_SCC_Abstract_2025_Homepage mit Copyright Note.pdf
Size
2.26 MB
Format
Adobe PDF