# 300-GHz Photonics-Aided Wireless 2 × 2 MIMO Transmission over 200 m Using GMM-Enhanced Duobinary Unsupervised Adaptive CNN

## TL;DR

A new wireless system using advanced signal processing achieves 100 Gbit/s data rates over 200 meters at 300 GHz, improving performance in challenging terahertz communication environments.

## Contribution

A label-free GMM-enhanced duobinary unsupervised adaptive CNN is introduced for robust terahertz wireless communication without training labels.

## Key findings

- The proposed DB-UACNN achieves up to 2.6 dB additional SNR gain over conventional duobinary filtering.
- The system maintains BER below the 7% HD-FEC threshold at 100 Gbit/s over 200 m in a 300-GHz MIMO setup.
- Duobinary shaping provides 1.87 dB and 1.70 dB SNR gain in X- and Y-polarization, respectively.

## Abstract

What are the main findings?
A GMM-enhanced duobinary unsupervised adaptive CNN (DB-UACNN) is proposed to enhance duobinary filtering without relying on training labels, enabling robust operation under unknown terahertz wireless channels.In a 300-GHz photonics-aided 2 × 2 MIMO wireless system over 200 m, the proposed scheme achieves up to 2.6 dB additional SNR gain over conventional duobinary filtering and maintains BER below the 7% HD-FEC threshold at 100 Gbit/s.

A GMM-enhanced duobinary unsupervised adaptive CNN (DB-UACNN) is proposed to enhance duobinary filtering without relying on training labels, enabling robust operation under unknown terahertz wireless channels.

In a 300-GHz photonics-aided 2 × 2 MIMO wireless system over 200 m, the proposed scheme achieves up to 2.6 dB additional SNR gain over conventional duobinary filtering and maintains BER below the 7% HD-FEC threshold at 100 Gbit/s.

What are the implications of the main findings?
The proposed label-free, GMM-guided learning framework removes the dependence on pilot symbols or ideal references, improving spectral efficiency and adaptability in practical THz wireless systems.Combining duobinary shaping, MLSD, and unsupervised neural enhancement offers a scalable DSP scheme for future high-capacity MIMO terahertz links, supporting beyond-100 Gbit/s transmission in bandwidth-limited and nonlinear environments.

The proposed label-free, GMM-guided learning framework removes the dependence on pilot symbols or ideal references, improving spectral efficiency and adaptability in practical THz wireless systems.

Combining duobinary shaping, MLSD, and unsupervised neural enhancement offers a scalable DSP scheme for future high-capacity MIMO terahertz links, supporting beyond-100 Gbit/s transmission in bandwidth-limited and nonlinear environments.

Terahertz wireless communication offers ultra-high bandwidth, enabling an extremely high data rate for next-generation networks. However, it faces challenges including severe propagation loss and atmospheric absorption, which limits the transmission rate and transmission distance. To address the problem, polarization division multiplexing (PDM) and antenna diversity techniques are utilized in this work to increase system capacity without changing the bandwidth of transmitted signals. Meanwhile, duobinary shaping is used to solve the problem of bandwidth limitation of components in the system, and the final duobinary signals are recovered by maximum likelihood sequence detection (MLSD). A Gaussian mixture model (GMM)-enhanced duobinary unsupervised adaptive convolutional neural network (DB-UACNN) is proposed, to further deal with channel noise. Based on the technologies above, a 2 × 2 multiple-input multiple-output (MIMO) photonic-aided terahertz wireless transmission system at 300 GHz is demonstrated. Experimental results have proved that the signal-to-noise ratio (SNR) gain of duobinary shaping is up to 1.87 dB and 1.70 dB in X-polarization and Y-polarization. The proposed GMM-enhanced DB-UACNN also shows extra SNR gain of up to 2.59 dB and 2.63 dB in X-polarization and Y-polarization, compared to the conventional duobinary filter. The high transmission rate of 100 Gbit/s over the distance of 200 m is finally realized under a 7% hard-decision forward error correction (HD-FEC) threshold.