Learning to Dynamically Allocate Radio Resources in Mobile 6G in-X Subnetworks

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

27 Downloads (Pure)

Abstrakt

This paper investigates efficient deep learning based methods for interference mitigation in independent wireless subnetworks via dynamic allocation of radio resources. Resource allocation is cast as a mapping from interference power measurements at each subnetwork to a class of shared frequency channels. A deep neural network (DNN) is then trained to approximate this mapping using data obtained via application of centralized graph coloring (CGC). The trained network is then deployed at each subnetwork for distributed channel selection. Simulation results in an environment with mobile subnetworks have shown that relatively small-sized DNNs can be trained offline to perform distributed channel allocation. The results also show that regardless of the choice of initialization, a DNN for distributed channel selection can achieve similar performance as CGC up to a probability of loop failure (PLF) of 6 × 10-5 in diverse environments with only aggregate interference power measurements as input.

OriginalsprogEngelsk
Titel2021 IEEE 32nd Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)
Antal sider7
ForlagIEEE
Publikationsdato16 sep. 2021
Sider959-965
ISBN (Trykt)978-1-7281-7587-4
ISBN (Elektronisk)978-1-7281-7586-7
DOI
StatusUdgivet - 16 sep. 2021
Begivenhed2021 IEEE 32nd Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) - Helsinki, Finland
Varighed: 13 sep. 202116 sep. 2021

Konference

Konference2021 IEEE 32nd Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC)
Land/OmrådeFinland
ByHelsinki
Periode13/09/202116/09/2021
NavnI E E E International Symposium Personal, Indoor and Mobile Radio Communications
ISSN2166-9570

Fingeraftryk

Dyk ned i forskningsemnerne om 'Learning to Dynamically Allocate Radio Resources in Mobile 6G in-X Subnetworks'. Sammen danner de et unikt fingeraftryk.

Citationsformater