Advanced terahertz neural network unveiled at City University of Hong Kong

HONG KONG (10 November 2024)— An innovative planar spoof plasmonic neural network (SPNN) platform capable of directly detecting and processing terahertz (THz) electromagnetic signals has been unveiled by researchers at City University of Hong Kong (CityUHK) and Southeast University in Nanjing.

The study has enormous potential for fields such as intelligent communication, advanced computing systems, and terahertz on-chip integration, all of which are crucial for the future of 6G.

The research project is led by Professor Chan Chi-hou, Chair Professor of the Department of Electrical Engineering and Director of State Key Laboratory of Terahertz and Millimeter Waves (SKLTMW) at CityUHK and Academician Cui Tiejun, Director of State Key Laboratory of Millimeter Waves, Southeast University.

The paper, “Terahertz spoof plasmonic neural network for diffractive information recognition and processing,” was recently published in Nature Communications.

The team set out to address the challenges posed by the rapid evolution of artificial intelligence. Traditional space-diffractive neural networks suffer from low-space transmission efficiency and large spatial dimensions, limiting their miniaturisation and broader applications. This new SPNN platform overcomes these limitations by offering a compact, efficient, and easily integrable solution.

The new technology, composed of compact spoof surface plasmon polaritons diffraction layers and phase-shifting layers, introduces a compact method for building and utilising neural networks. It can efficiently handle complex tasks like handwriting recognition and multi-user distinction, offering potential applications in terahertz on-chip integration and intelligent communication systems.

“The SPNN can directly process different users’ directions on the THz platform, integrating the capability of classifying handwritten digits without relying on digital processing,” said Dr Gao Xinxin, the first author of the paper and a postdoctoral fellow at SKLTMW.

The SPNN’s compactness, efficiency, and scalability make it an ideal candidate for artificial neural networks, addressing the power consumption and scalability issues of traditional digital computers. This network can directly process and recognise diffractive information with low power consumption and at the speed of light, broadening the application of terahertz plasmonic metamaterials.

“SKLTMW has excellent fabrication and measurement facilities supported by the Research Grants Council, the Innovation and Technology Commission of the HKSAR Government, and CityUHK,” said Professor Chan. “These facilities allow us to test our ideas promptly and generate unexpected results.”

Gu Ze and Dr Ma Qian, a PhD student and postdoctoral fellow, respectively, at Southeast University, are co-first authors of the paper. Other contributors are Cui Wenyi, PhD student, Professor You Jianwei of Southeast University, and Dr Chen Baojie and Dr Shum Kam-man of SKLTMW. Dr Ma, Academician Cui, and Professor Chan are the corresponding authors.