Related Experiment Video
Updated: May 23, 2025
05:57
Characterization of SiN Integrated Optical Phased Arrays on a Wafer-Scale Test Station
Published on: April 1, 2020
8.0K
Near-Sensor Edge Computing System Enabled by a CMOS Compatible Photonic Integrated Circuit Platform Using Bilayer
Zhihao Ren1,2,3, Zixuan Zhang1,2, Yangyang Zhuge1,2
1Department of Electrical and Computer Engineering, National University of Singapore, Singapore, 117583, Singapore.
Nano-Micro Letters
|May 19, 2025
Summary
This study presents a near-sensor edge computing (NSEC) system for real-time, energy-efficient artificial intelligence (AI) at the edge. The novel AlN/Si waveguide platform enables high-accuracy AI for applications like healthcare and robotics.
Area of Science:
- Photonics and Artificial Intelligence (AI)
- Edge Computing Hardware
Background:
- Large-scale AI models require significant computational power and cloud infrastructure, leading to latency, privacy, and energy concerns.
- Existing AI hardware solutions face limitations in real-world, edge applications like wearable health monitoring and robotics.
Purpose of the Study:
- To develop an innovative near-sensor edge computing (NSEC) system for real-time, energy-efficient AI capabilities at the edge.
- To address the limitations of centralized AI by bringing computation closer to data acquisition.
Main Methods:
- A bilayer aluminum nitride (AlN)/silicon (Si) waveguide platform was utilized for the NSEC system.
- Photonic feature extraction was achieved using AlN microring resonators, while Si-based thermo-optic Mach-Zehnder interferometers performed neural network computations.
Main Results:
- The NSEC system demonstrated high classification accuracies: 96.77% for gesture analysis and 98.31% for gait analysis.
- Ultra-low latency (< 10 ns) and minimal energy consumption (< 0.34 pJ) were achieved.
- Successful multimodal gesture and gait analysis was performed at the edge.
Conclusions:
- The developed NSEC system offers a transformative approach to AI hardware, enabling efficient and privacy-preserving AI solutions.
- This advancement bridges the gap between AI models and real-world applications, paving the way for next-generation human-machine interfaces.
- The system marks a pivotal advancement in edge computing and AI deployment for diverse fields including healthcare and robotics.
