Cheung, Hewlett Packard Enterprise Researchers Published in Nature Communications

NC State researchers have demonstrated a new method for searching and storing information in a system using semiconductor lasers and silicon photonics. Practical applications include improving the speed and efficiency of AI and machine learning.

NC State ECE Associate Professor Stan Cheung and researchers from Hewlett Packard Enterprise (HPE) recently published a paper, “Heterogeneous III-V/Si micro-ring laser array with multi-state non-volatile memory for ternary content-addressable memories” in Nature Communications.

“This publication in Nature Communications is exciting because it really opens up the field for others to push this even further.”

Current systems that connect optical networks with traditional electronic memory typically need extra steps to convert signals, which slows down a system. To address this problem, the research team demonstrated the use of non-volatile lasers for optical ternary content-addressable memories (O-TCAM); applications include fast memory search and in-memory computing functions necessary for a variety of machine learning algorithms.

Essentially, the team has used semiconductor lasers and silicon photonic ring modulators, roughly the diameter of a hair, to create a type of memory that finds data by its content, not location. This new, super fast method of searching and storing information can revolutionize how AI, machine learning and other data-intensive applications process information.

“These results validate the feasibility of using this technology for realizing larger-scale CAMs that can operate at higher speeds than electrical counterparts,” said Cheung. “This publication in Nature Communications is exciting because it really opens up the field for others to push this even further.”

Visit the department’s LinkedIn page to learn more about the research.