100-Gb/s All-Electronic THz Wireless Direct Detection System Enabled by Kramers--Kronig Inspired Neural Network
IEEE Transactions on Microwave Theory and Techniques · 2026-07 · 74 (7) · 6638-6649
DOI: 10.1109/tmtt.2026.3687932 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Microwave Theory and Techniques, 2026-07. Verfasst von Zhuo-Yue Wen und 9 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 33 Literaturangaben. Gefördert von: China Postdoctoral Science Foundation, Mobile Information Networks-National Science and Technology Major Project.
Abstract (Originalsprache)
The Kramers–Kronig (KK) receiver paradigm offers a cost-effective and simplified solution for short-range high-speed direct detection (DD) transmission in future 6G wireless networks. Currently, high-speed terahertz (THz) KK technology demonstrations predominantly rely on photonics-assisted techniques, whereas demonstrations based on integration-friendly all-electronic architectures are scarce. All-electronic THz KK systems based on harmonic mixing transmitters require not only strict hardware-level sideband filtering to satisfy the single-sideband minimum phase (SSB-MP) condition but also robust algorithms to mitigate the complex mixed impairments inherent to the THz links. Here, a nonlinearity-robust equalization and SSB-MP signal recovery fusion scheme based on KK-inspired single-hidden-layer neural networks (SHL-NNs) is proposed. Unlike deep learning, the proposed architecture is heuristically designed to fuse the mathematical operators of the KK relation and Volterra series equalization into a unified shallow topology. This design enables the reconstruction of the SSB-MP signal, while simultaneously mitigating frequency-selective fading and transceiver nonlinearity in THz links. Numerical simulations demonstrate that the proposed scheme achieves a 5.8 and 1.5 dB improvement in error vector magnitude (EVM) compared to conventional KK (CKK) and generalized KK (GKK) algorithms, respectively. Validated experimentally in a G-band ( $165\sim 200$ GHz) all-electronic wireless link, the scheme supports a line rate of 100 Gb/s at a 25% soft-decision forward error correction (SD-FEC) threshold using 16-QAM modulation by effectively handling the filtered MP signal and compensating for hardware impairments. This work experimentally validates that the all-electronic THz DD system, enabled by the KK-inspired neural network (NN), can support 100-Gb/s-class transmission.
Förderung
- China Postdoctoral Science Foundation
- Mobile Information Networks-National Science and Technology Major Project
Zitieren
APA
Wen, Z., Zhang, Q., Han, F., Liang, Z., Huang, Z., Chen, Y., Fan, L., Yu, X., Li, W., & Tao, X. (2026). 100-Gb/s All-Electronic THz Wireless Direct Detection System Enabled by Kramers--Kronig Inspired Neural Network. IEEE Transactions on Microwave Theory and Techniques, 74(7), 6638-6649. https://doi.org/10.1109/tmtt.2026.3687932
Vancouver
Wen Z, Zhang Q, Han F, Liang Z, Huang Z, Chen Y, et al. 100-Gb/s All-Electronic THz Wireless Direct Detection System Enabled by Kramers--Kronig Inspired Neural Network. IEEE Transactions on Microwave Theory and Techniques. 2026;74(7):6638-6649. doi:10.1109/tmtt.2026.3687932
BibTeX
@article{Wen2026,
author = {Wen, Zhuo-Yue and Zhang, Qun and Han, Feifan and Liang, Zhu and Huang, Zebin and Chen, Yuxuan and Fan, Linsheng and Yu, Xiongbin and Li, Weichao and Tao, Xiaofeng},
title = {100-Gb/s All-Electronic THz Wireless Direct Detection System Enabled by Kramers--Kronig Inspired Neural Network},
journal = {IEEE Transactions on Microwave Theory and Techniques},
year = {2026},
volume = {74},
number = {7},
pages = {6638--6649},
doi = {10.1109/tmtt.2026.3687932}
}
EUVIMED