Deep Learning Model for CS-Based Signal Recovery for IRS-Assisted Near-Field THz MIMO System

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F29142890%3A_____%2F24%3A00048979" target="_blank" >RIV/29142890:_____/24:00048979 - isvavai.cz</a>

Výsledek na webu

<a href="https://ieeexplore.ieee.org/document/10660298" target="_blank" >https://ieeexplore.ieee.org/document/10660298</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/OJVT.2024.3452412" target="_blank" >10.1109/OJVT.2024.3452412</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Deep Learning Model for CS-Based Signal Recovery for IRS-Assisted Near-Field THz MIMO System

Popis výsledku v původním jazyce

Terahertz (THz) communication is a cutting-edge technology for the sixth-generation (6G) networks, offering vast bandwidths and data rates up to terabits per second, significantly advancing vehicular connectivity and services. However, THz signals are impacted by attenuation, path loss, and beam misalignment. Furthermore, the requisite high Nyquist sampling rates for THz systems increase the computational and system complexity of the receiver. A promising solution to navigate these obstacles involves the use of intelligent reflecting surfaces (IRS)-enhanced multiple-input multiple-output (MIMO) technology, which steers THz wave propagation. However, the substantial dimensions associated with IRS and MIMO extend the near-field, particularly at THz frequencies, as indicated by the Rayleigh distance and suffer from beam squint. To reduce system complexity and reduce sampling to sub-Nyquist rate, we propose a novel receiver design for an IRS-assisted near-field MIMO THz system that employs low-complexity compressed sensing. This method introduces an IRS signal-matched (IRSSM) measurement matrix with beam squint for capturing the transmitted signal at a sub-Nyquist rate, taking advantage of the sparsity in the signal and THz channels, and signal recovery using the deep learning (DL) model. Simulation results for symbol error rate (SER) and normalized mean square error (NMSE) performance indicate that the proposed DL-based receiver outperforms conventional recovery algorithms based on orthogonal matching pursuit (OMP) CS-recovery and dictionary-shrinkage estimation (DSE).

Název v anglickém jazyce

Deep Learning Model for CS-Based Signal Recovery for IRS-Assisted Near-Field THz MIMO System

Popis výsledku anglicky

Terahertz (THz) communication is a cutting-edge technology for the sixth-generation (6G) networks, offering vast bandwidths and data rates up to terabits per second, significantly advancing vehicular connectivity and services. However, THz signals are impacted by attenuation, path loss, and beam misalignment. Furthermore, the requisite high Nyquist sampling rates for THz systems increase the computational and system complexity of the receiver. A promising solution to navigate these obstacles involves the use of intelligent reflecting surfaces (IRS)-enhanced multiple-input multiple-output (MIMO) technology, which steers THz wave propagation. However, the substantial dimensions associated with IRS and MIMO extend the near-field, particularly at THz frequencies, as indicated by the Rayleigh distance and suffer from beam squint. To reduce system complexity and reduce sampling to sub-Nyquist rate, we propose a novel receiver design for an IRS-assisted near-field MIMO THz system that employs low-complexity compressed sensing. This method introduces an IRS signal-matched (IRSSM) measurement matrix with beam squint for capturing the transmitted signal at a sub-Nyquist rate, taking advantage of the sparsity in the signal and THz channels, and signal recovery using the deep learning (DL) model. Simulation results for symbol error rate (SER) and normalized mean square error (NMSE) performance indicate that the proposed DL-based receiver outperforms conventional recovery algorithms based on orthogonal matching pursuit (OMP) CS-recovery and dictionary-shrinkage estimation (DSE).

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20200 - Electrical engineering, Electronic engineering, Information engineering

Projekt

—

Návaznosti

N - Vyzkumna aktivita podporovana z neverejnych zdroju

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

IEEE Open Journal of Vehicular Technology

ISSN

2644-1330

e-ISSN

—

Svazek periodika

5

Číslo periodika v rámci svazku

1

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

10

Strana od-do

1-10

Kód UT WoS článku

001327401100002

EID výsledku v databázi Scopus

—