Data-Driven Estimation of Capacity Upper Bounds
Artikel i vetenskaplig tidskrift, 2022

We consider the problem of estimating an upper bound on the capacity of a memoryless channel with unknown channel law and continuous output alphabet. A novel data-driven algorithm is proposed that exploits the dual representation of capacity where the maximization over the input distribution is replaced with a minimization over a reference distribution on the channel output. To efficiently compute the required divergence maximization between the conditional channel and the reference distribution, we use a modified mutual information neural estimator that takes the channel input as an additional parameter. We numerically evaluate our approach on different memoryless channels and show empirically that the estimated upper bounds closely converge either to the channel capacity or to best-known lower bounds.

Artificial neural networks

channel capacity

upper capacity bounds

Channel estimation

mutual information neural estimation

Training

duality

Autoencoders

divergence estimation

Estimation

dual capacity representation

neural networks

Upper bound

Neurons

Mutual information

Författare

Christian Häger

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

Erik Agrell

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

IEEE Communications Letters

1089-7798 (ISSN) 15582558 (eISSN)

Vol. 26

Fysikbaserad djupinlärning för optisk dataöverföring och distribuerad avkänning

Vetenskapsrådet (VR) (2020-04718), 2021-01-01 -- 2024-12-31.

Kommunikation över optiska kanaler med skurfel

Vetenskapsrådet (VR) (2021-03709), 2022-01-01 -- 2025-12-31.

Frigöra full fiberoptisk kapacitet

Knut och Alice Wallenbergs Stiftelse (KAW 2018.0090), 2019-07-01 -- 2024-06-30.

Styrkeområden

Informations- och kommunikationsteknik

Ämneskategorier

Telekommunikation

Kommunikationssystem

Signalbehandling

Infrastruktur

C3SE (Chalmers Centre for Computational Science and Engineering)

DOI

10.1109/LCOMM.2022.3207385

Mer information

Senast uppdaterat

2024-07-17