Neural-network-based carrier-less amplitude phase modulated signal generation and end-to-end optimization for fiber-terahertz integrated communication system

Optics Express, Год журнала: 2024, Номер 32(6), С. 8623 - 8623

Опубликована: Фев. 15, 2024

In fiber-terahertz integrated communication systems, nonlinear distortion and inter-symbol interference (ISI) will degrade transmission performance. Pre-compensation is an efficient method to handle the channel as it can avoid noise boosting during compensation reduce receiver side signal processing algorithmic complexity at user-end (UE) considering asymmetric access scenario. this paper, we propose experimentally demonstrate a neural-network (NN)-based carrier-less amplitude phase (CAP) modulated generation end-to-end optimization for system. The CAP generated directly from quadrature modulation symbols pre-compensated through transmitter NN, which allows demodulate with simple linear digital process (DSP). generating signal, NN based learns group of filters, generate, up-convert, pre-compensate signals. Based on proposed method, integration system 220 GHz demonstrated sensitivity gain 1.2 dB achieved speed 50 Gbps forward error correction (FEC) bit rate (BER) threshold 1 × 10

Язык: Английский

---

Leveraging autoencoder for laser phase noise compensation in coherent optical OFDM systems

Journal of Optics, Год журнала: 2025, Номер unknown

Опубликована: Янв. 29, 2025

Язык: Английский

---

基于端到端深度学习的相位噪声补偿方案研究

Acta Optica Sinica, Год журнала: 2025, Номер 45(9), С. 0906004 - 0906004

Опубликована: Янв. 1, 2025

---

Recent Advances on Machine Learning-aided DSP for Short-reach and Long-haul Optical Communications

Optical Fiber Communication Conference (OFC) 2022, Год журнала: 2025, Номер unknown, С. W4H.1 - W4H.1

Опубликована: Янв. 1, 2025

In this paper, we highlight recent advances in the use of machine learning for implementing equalizers optical communications. We both algorithmic as well implementation aspects using conventional and neuromorphic hardware.

Язык: Английский

---

Performance investigation of geometric constellation shaping-based coherent WDM optical fiber communication system supported by deep-learning autoencoder

Results in Optics, Год журнала: 2024, Номер 15, С. 100629 - 100629

Опубликована: Фев. 9, 2024

Geometric constellation shaping (GCS) has been proposed to enhance the performance of wavelength-division multiplexing (WDM) coherent optical fiber communication (OFC) systems. This paper presents a comprehensive simulation platform design and simulate these systems' using an advanced autoencoder (AE) technique. The AE is based on end-to-end deep learning takes into account system parameters, especially those related OFC channel. uses identical AEs, one in each WDM channel, only central channel trained achieve required target. developed capable assessing bit error rate characteristics, signal-to-noise ratio, mutual information for long-haul OFC-WDM systems operating with various modulation formats. results indicate that multilayer perceptron neural network requires four hidden layers, sixteen nodes per layer, "16x order" batch size optimum performance. AE's behaviour investigated identify allowed ranges optimal launch power two different procedures better 2D 4D constellations.

Язык: Английский

---

Hybrid Constellation Shaping 64QAM Based on Hexagonal Lattice of Constellation Subset

Photonics, Год журнала: 2023, Номер 10(9), С. 1008 - 1008

Опубликована: Сен. 4, 2023

Increasing demand for higher-speed and large-capacity data communications has driven the development of constellation shaping technology. This paper proposes a hybrid scheme 64-quadrature amplitude modulation (64QAM) based on hexagonal lattice subset. The proposed aims to enhance nonlinear tolerance higher-order modulated signals further improve gain. initial quantitative characterization is firstly performed structure. Then, objective function maximizing figure merits (CFM) utilized determine position distribution points, resulting in generation geometric shaping-64QAM (GS-64QAM) signal. Finally, according concentric layers, all points are divided into multiple subsets where within same subset assigned probability, (HS-64QAM) signal generated. To validate effectiveness scheme, experimental verification was demonstrated 120 Gbit/s multi-span coherent optical communication system. Experimental results indicate that, at soft-decision forward error correction threshold, HS-64QAM achieves an signal-to-noise ratio (OSNR) gain 1.9 dB 4.1 over uniform GS-64QAM back-to-back 375 km transmission scenarios, respectively. Furthermore, OSNR 2.7 7.6 Square-64QAM

Язык: Английский

---

Autoencoder Learning of Constellation Shaping Robust to Semiconductor Laser Noise and Nonlinearity in Fiber-THz System

Optical Fiber Communication Conference (OFC) 2022, Год журнала: 2024, Номер unknown, С. M4K.5 - M4K.5

Опубликована: Янв. 1, 2024

We experimentally demonstrate the robustness of autoencoder-based constellation shaping against semiconductor laser noise and nonlinearity. Up to 46% lower BER 1.5 dB gain are achieved in fiber-THz system at 320 GHz.

Язык: Английский

---

High-Capacity Coherent WDM Networks Empowered by Probabilistic Shaping and End-to-End Deep Learning

Journal of Telecommunications and Information Technology, Год журнала: 2024, Номер unknown, С. 71 - 81

Опубликована: Июнь 4, 2024

To optimize the functionality of coherent optical fiber communication (OFC) systems and enhance their capacity related to long-haul transmissions, wavelength-division multiplexing (WDM) probabilistic constellation shaping (PCS) techniques have been used. This paper develops an end-to-end (E2E) deep learning (DL)-based PCS algorithm, i.e., autoencoder (AE) for a high-order modulation format WDM system that minimizes nonlinear effects while ensuring high considers parameters, in particular those OFC channel. Only AE central channel is trained meet specified performance objective, as design employs identical AEs each The simulation results show architecture should consist two hidden layers, with thirty nodes per layer ”32×modulation order” batch size obtain optimal performance, when designing using dense neural network. behavior examined determine optimum launch-power ranges system's performance. developed AE-based PCS-WDM provides 0.4 gain outperforms conventional solutions.

Язык: Английский

---

Nonlinearity-Aware End-to-End Learning Architecture for Next Generation Wireless Backhaul

2022 IEEE Wireless Communications and Networking Conference (WCNC), Год журнала: 2024, Номер unknown, С. 1 - 6

Опубликована: Апрель 21, 2024

Язык: Английский

---

$M$th Power Carrier Phase Estimation with Wiener Phase Noise for $M\text{PSK}$ Modulations

2022 IEEE Wireless Communications and Networking Conference (WCNC), Год журнала: 2024, Номер unknown, С. 1 - 6

Опубликована: Апрель 21, 2024

Язык: Английский

---