Chinese Researchers Report 3.2 Pb/s Data Transfer Over 24-Core Fiber

Researchers in China say they have pushed data through a 24-core optical fiber at 3.2 petabits per second, a figure they describe as a new real-time transmission record, according to TechRadar. The team says the link could theoretically carry the equivalent of thousands of ultra-high-definition films in a single second.

The result tops the group's own earlier mark of 2.5 Pb/s, which used a similar 24-core fiber design and was reported earlier this year. The work involved staff from several Chinese institutions, including a state optical communications laboratory, a telecoms technology group, Pengcheng Laboratory and a fiber manufacturing joint venture.

The researchers combined two approaches meant to squeeze more capacity out of optical networks. One splits traffic into separate channels inside a single fiber so several streams can run at once, while the other spreads signals across three wavelength ranges known as S, C and L bands, covering a broader slice of the optical spectrum.

Merging the two methods created engineering problems. Signals travelling side by side interfered with one another, and the three bands behaved differently, so the team had to keep those variations from degrading overall performance.

The test ran continuously in real time on hardware the team describes as meeting commercial requirements. That sets it apart from trials that lean on bespoke lab gear or signal processing built only for research.

The Hubei provincial government website likened the capacity to 3.2 million households each running a separate 1 Gbps connection at the same moment. It also said the rate could theoretically pull more than 20,000 4K movies in one second under ideal conditions. Those figures describe network capacity in principle, not speeds any ordinary broadband customer would receive.

The figure is not the fastest reported over a 24-core fiber. Lab work has hit 5.29 Pb/s with similar multi-core cable, more than two petabits above this result, though such experiments may rely on specialist equipment and processing unsuited to everyday networks.

Commercial-grade hardware is a notable part of the claim, but it does not mean the technology is ready to sell. A single live trial cannot show long-term dependability, affordable running costs or compatibility with existing telecom infrastructure under heavy load. No deployment timetable, pricing or independent assessment of sustained performance has been published.