Scholarlink

<p>The exponential growth of data capacity in intelligent terminals drives higher data traffic toward network edges. Compact I/O systems are essential to support space-constrained infrastructures at the computing edges or modular data centers. However, scaling high-capacity transmission via increasing physical channels is constrained by limited source coherence and low carrier-to-noise ratios (OCNR), hindering lightweight, efficient applications like distributed edge computing. Here, we exploit an integrated self-injection-locked dark-pulse microcomb to achieve 1 Tbps/λ/core transmission and characterize the constraints among OCNR, linewidth, and transmission rate. Furthermore, a multi-dimensional transmission architecture for multi-nodes aggregation is explored, boosting the transmission rate to 200 Tbps with 16 comblines at 70 Gbaud. Combining with integrated waveshapers and semiconductor optical amplifiers, a chip-level parallel carrier generator is explored, reducing system size a hundredfold while delivering 5 Tbps. Our results highlight significant potential for compact and resource-conserving transmission systems in data centers and distributed high performance computing applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploiting a centrally powered coherent microcomb for lightweight optical transmission

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月29日

查看全文
  • Junhao Han,
  • Guofeng Yan,
  • Kang Li,
  • Bitao Shen,
  • Haowen Shu,
  • Yimeng Wang,
  • Yuchen Zhang,
  • Jiong Xiao,
  • Yichen Wu,
  • Huajin Chang,
  • Chengkun Cai,
  • Xuguang Zhang,
  • Xingjun Wang,
  • Jian Wang

摘要

The exponential growth of data capacity in intelligent terminals drives higher data traffic toward network edges. Compact I/O systems are essential to support space-constrained infrastructures at the computing edges or modular data centers. However, scaling high-capacity transmission via increasing physical channels is constrained by limited source coherence and low carrier-to-noise ratios (OCNR), hindering lightweight, efficient applications like distributed edge computing. Here, we exploit an integrated self-injection-locked dark-pulse microcomb to achieve 1 Tbps/λ/core transmission and characterize the constraints among OCNR, linewidth, and transmission rate. Furthermore, a multi-dimensional transmission architecture for multi-nodes aggregation is explored, boosting the transmission rate to 200 Tbps with 16 comblines at 70 Gbaud. Combining with integrated waveshapers and semiconductor optical amplifiers, a chip-level parallel carrier generator is explored, reducing system size a hundredfold while delivering 5 Tbps. Our results highlight significant potential for compact and resource-conserving transmission systems in data centers and distributed high performance computing applications.