Related Experiment Video
Updated: Jul 11, 2025
05:57
Characterization of SiN Integrated Optical Phased Arrays on a Wafer-Scale Test Station
Published on: April 1, 2020
8.1K
Deep neural network-based phase calibration in integrated optical phased arrays.
Jae-Yong Kim1, Junhyeong Kim1, Jinhyeong Yoon1
1School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 34141, Republic of Korea.
Scientific Reports
|November 15, 2023
Summary
A novel deep neural network (DNN) method calibrates phase errors in optical phased arrays (OPAs) for improved beamforming. This intelligent approach efficiently manages complex phase control across 64 channels.
Area of Science:
- Photonics and Optical Engineering
- Artificial Intelligence in Optics
- Integrated Optical Systems
Background:
- Phase calibration is critical for accurate beamforming in integrated optical phased arrays (OPAs).
- Traditional methods struggle with the complexity of multi-channel phase control.
- Phase errors degrade beamforming performance in OPAs.
Purpose of the Study:
- To introduce a novel deep neural network (DNN) based phase calibration methodology for integrated OPAs.
- To enhance beamforming precision and efficiency in 64-channel OPAs.
- To address the complexity of voltage set combinations for precise phase control.
Main Methods:
- Development of a tandem deep neural network (DNN) architecture for phase calibration.
- Individual phase control for each of the 64 OPA channels using electro-optic phase shifters.
- Optimization through selective data sorting and hyperparameter tuning.
- Comparison of DNN-predicted beams against reference beams from the hill climbing algorithm.
Main Results:
- The DNN model achieved substantial agreement with reference beams, demonstrating practical effectiveness.
- A minor average intensity reduction of 0.84 dB in peak beam values was observed.
- A slight decrease of 0.06 dB in the side-mode-suppression-ratio was noted.
Conclusions:
- The DNN-based phase calibration methodology is practically effective for integrated OPA beamforming.
- The approach offers intelligent and time-efficient calibration for complex multi-beam scenarios.
- This method shows potential for advancing beamforming capabilities in optical systems.
