Finding a New Path for High-Speed Data Connections

[Labs_Editorial]

One of the primary ways researchers are feeding the compute demands of modern apps like generative AI is by using optical technologies to create faster data connections. They’re making progress. But, until recently, they’ve struggled to overcome a significant hurdle. It’s been hard to transmit information at high bandwidths without losing efficiency in the device’s performance.

Labs researchers are proposing a new way to get around this conundrum. They’ve come up with a new modulation scheme that reduces parasitic losses.

The proposal is detailed in a new research paper published Jan. 31 in Nature Communications. The paper, called “A 5×200 Gbps DWDM Microring Modulator Silicon Chip Empowered by Two-Segment Z-Shape Junctions,” was prepared by a team in the Large-Scale Silicon Photonics Lab led by Labs research scientist Yuan Yuan.

Yuan, who has worked at Hewlett Packard Labs for four years, said he hopes the innovation translates into speed and productivity improvements in high-performance computers.

“When you tackle a cutting-edge problem, it’s all about solving the engineering tradeoff,” he said. “On the one hand you want to push engineering efficiency to save power, but on the other you want it to be faster than before. Our challenge was to find the right recipe and the right combination that allowed us to do something others cannot do.”

Yuan’s team tested two concepts. First, they improved the efficiency of the silicon microring resonator modulators – the mechanism that modulates light. They did this by changing the distribution of the silicon-based impurities that guide the electrical charges inside the wave guide of a modulator. Directing the modulation charges through new paths – defined by Z-shaped junctions – reduced the device’s losses.

Second, they used a technique called five-channel wavelength division multiplexing that sends signals along five wavelengths of light – creating higher-bandwidth transmissions.

“We reached the level of 1 TB per second, which is the next generation of internet and other types of protocols that are used in supercomputers,” Yuan said. “We’re going to need this type of speed in the future.”

The paper was written by Yuan Yuan, along with other Labs researchers Yiwei Peng, Wayne V. Sorin, Stanley Cheung, Zhihong Huang, Di Liang, Marco Fiorentino, and Raymond G. Beausoleil.