12 Mai Towards the 3.2T optical interface
The companies Keysight Technologies, NTT Innovative Devices and Lumentum Holdings have jointly announced the demonstration of 448 Gbps data transmission using 224 GBaud PAM4 optical technology. According to the partners, this collaboration will address the increasing bandwidth requirements of artificial intelligence (AI) and machine learning (ML) applications and enable the development of energy-efficient 3.2-terabit interfaces for future data center networks.
AI and ML applications require ultra-fast data processing in real time. The transmission speed of 448 Gbit/s accelerates data transfer between computing and network nodes, reduces latency and thus also improves the efficiency of AI/ML models. As these models become increasingly complex, higher data throughput is essential to support scalability without bottlenecks. In addition, this speed boost should increase data center efficiency, reduce operating costs and minimize environmental impact.
Keysight says it is supporting the industry in the early R&D phase of 448 Gbit/s. During this phase, experts are researching various modulation methods, including PAM4, PAM6 and PAM8. For optical transmission at 224 Gbit/s, the industry has already chosen PAM4 as the preferred method, according to the partners.
The demonstration will reportedly use two Keysight M8199B arbitrary waveform generators with NTT’s frequency domain interleaver to generate 448 Gbps electrical signals and 224 Gbaud PAM4. This signal is used to drive an externally modulated Lumentum indium phosphide laser (EML). The successful generation of 240 Gbaud PAM4 is also demonstrated using the same hardware.
“Achieving 448 Gbps is a significant milestone in the advancement of optical interconnects for AI-driven cloud infrastructures,” said Dr. Matthew Sysak, Chief Technology Officer, Cloud and Networking Platform at Lumentum. “This collaboration underscores Lumentum’s expertise in photonics and our commitment to delivering cutting-edge optical components that support the rapid expansion of AI and cloud data centers.”
Source and image: www.keysight.com