EUVIMEDEUVIMED entdecken
Crossref

Artificial intelligence enhanced chip-on-board hybrid-packaged photonic-eletric silicon transceiver for high-speed data center interconnects

Jun Qin, Yu Sun, Qipeng Yang, Changhao Han, Yan Zhou, Siming Liu, Junde Lü, Yunhao Zhang, Yichen Wu, Yueqin Li, Jian Sun, Weiwei Hu und 5 weitere

Chinese Optics Letters · 2026 · 24 (7) · 072501

DOI: 10.3788/col202624.072501 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Chinese Optics Letters, 2026. Verfasst von Jun Qin und 16 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In this paper, a high-speed chip-on-board hybrid-packaged photonic-eletric silicon transceiver aided by deep neural network equalization is demonstrated. A four-channel silicon photonics Mach-Zehnder modulator with interleaved PN junctions and a photonics photodetector chip is seamlessly integrated with a four-channel driver chip and transimpedance amplifier (TIA) chip by chip-on-board hybrid-packaging. Simultaneously, an artificial fully connected multilevel real-valued deep neural network (MR-DNN) equalizer is proposed and implemented to equalize the linear and nonlinear distortions induced by the electrical and optical components in the photonic-eletric transceiver. Leveraging the deployed fully connected MR-DNN equalizer, a bit error rate (BER) below the soft decision forward error correction (SD-FEC) threshold can be achieved at a data rate of 160 Gbps and 120 Gbps PAM4 for the transmitter and receiver, respectively. By employing a higher-bandwidth Si slow-light modulator, the MR-DNN algorithm can easily accelerate the modulation to data rates higher than 224 Gbps and 260 Gbps. Experimental results indicate that the MR-DNN achieves superior performance compared to decision-directed least mean square (LMS), Volterra, and MR-ANN equalizers. The results show the potential of artificial-intelligence-based equalization techniques in future terabit-scale photonic-electric integrated silicon photonics

Zitieren

APA

Qin, J., Sun, Y., Yang, Q., Han, C., Zhou, Y., Liu, S., Lü, J., Zhang, Y., Wu, Y., Li, Y., Sun, J., Hu, W., He, Z., Yu, S., Wang, L., Wang, X., & Shu, H. (2026). Artificial intelligence enhanced chip-on-board hybrid-packaged photonic-eletric silicon transceiver for high-speed data center interconnects. Chinese Optics Letters, 24(7), 072501. https://doi.org/10.3788/col202624.072501

Vancouver

Qin J, Sun Y, Yang Q, Han C, Zhou Y, Liu S, et al. Artificial intelligence enhanced chip-on-board hybrid-packaged photonic-eletric silicon transceiver for high-speed data center interconnects. Chinese Optics Letters. 2026;24(7):072501. doi:10.3788/col202624.072501

BibTeX
@article{Qin2026,
  author = {Qin, Jun and Sun, Yu and Yang, Qipeng and Han, Changhao and Zhou, Yan and Liu, Siming and Lü, Junde and Zhang, Yunhao and Wu, Yichen and Li, Yueqin and Sun, Jian and Hu, Weiwei and He, Zhixue and Yu, Shaohua and Wang, Lei and Wang, Xingjun and Shu, Haowen},
  title = {Artificial intelligence enhanced chip-on-board hybrid-packaged photonic-eletric silicon transceiver for high-speed data center interconnects},
  journal = {Chinese Optics Letters},
  year = {2026},
  volume = {24},
  number = {7},
  pages = {072501},
  doi = {10.3788/col202624.072501}
}
Kostenlos starten →