Abstract:We propose a novel end-to-end (E2E) fiber-terahertz (THz) integrated system based on a data-driven mutual information neural estimator (MINE). This system utilizes the MINE neural networks for Tx-NNs optimization, enabling constellation shaping, channel sampling without known channel models, and a differentiable carrier phase estimation for resisting phase noise. Our approach follows state-of-the-art bitwise autoencoders (AE), which requires a differentiable implementation of all operations between transmitter and receiver, including the DSP algorithms. Firstly, to address the demand for higher spectral efficiencies, we employ the E2E-AE framework for constellation shaping in the presence of laser noises and nonlinearity. Subsequently, we rigorously examine the characterization of lasers with white and colored frequency modulation noise and investigate their impact on MI performance using simulation and experiment when the carrier phase recovery algorithm is present and absent. Finally, We experimentally evaluate our proposed E2E framework within a dual-polarized fiber-THz integrated system at 320GHz, considering linewidth, flicker ( $1/f$ ) noise, and resonance peaks. By joint optimization constellation shaping and compensation for the phase noise and nonlinearity, we achieve a robust and pilot-free modulation scheme improving the performances by up to 0.15-bit and 1.21-dB gains in MI and SNR compared to Grid-QAM constellations. Additionally, under the constraints of laser phase noise and nonlinearity, we achieve a transmission speed of 230Gbit/s in the E2E-AE fiber-THz system for the first time. This solution advances the E2E-AE framework by incorporating differentiable functions tailored for practical scenarios.

End-to-End Optimization of Directly Modulated Laser Links using Chirp-Aware Modeling

End-to-end optimization of optical communication systems based on directly modulated lasers

Experimental Demonstration of End-to-End Optimization for Directly Modulated Laser-based IM/DD Systems

Data-Driven Modeling of Directly-Modulated Lasers

Optimization of Inter-satellite Laser Communication Links

Direct Modulation of Deep-Submicron Slotted Single Mode FP Laser

Differentiable Machine Learning-Based Modeling for Directly-Modulated Lasers

200 Gb/s Unamplified IM/DD Transmission over 20-km SMF with an O-band Low-Chirp Directly Modulated Laser

Chirp-Managed Directly-Modulated DFB Laser

10Gb/s Direct Modulation of Widely Tunable V-Cavity-Laser with Chirp Managed Laser Technology

Model-based Deep Learning for End-to-end Optimization in the Fiber-Terahertz Communication System

Generation and Transmission of 10-Gb/s Rz-Dpsk Signals Using A Directly Modulated Chirp-Managed Laser

Chirp-Managed 42.8 Gbit/s Transmission over 20 Km Standard SMF Without DCF Using Directly Modulated Laser

Design, Optimization, and Blackbox Optimization of Laser Systems

Performances Improvement in Radio over Fiber Link Through Carrier Suppression Using Stimulated Brillouin Scattering

Electro-Assisted Optical Method of Chirp Compensation for Bandwidth/Bitrate Variable Transceivers Using the Hammerstein-Wiener System

Volterra Equalization of Complex Modulation Utilizing Frequency Chirp in Directly Modulated Lasers

End-to-End Optimization of Constellation Shaping Robust to Semiconductor Laser Noise and Nonlinearity in Fiber-THz System

Piecewise Linear and Nonlinearity-aware Complex-valued Decision Feedback Equalization for DML-DD Link at C Band

Multi-objective Optimization of Amplifier Operating Point and Launch Signal Power Pre-emphasis in a Cascade of EDFAs