| Title |
Research on Intelligent Automatic Control of the Bias Voltage in Mach-Zehnder Optical Modulators using Deep Neural Network-based Optimization Technique |
| Authors |
최동민(Dongmin Choi) ; 원용욱(Yong-Yuk Won) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.4.16 |
| Keywords |
Optical modulation; Auto bias control; MZM; Deep learning; Loss function optimization |
| Abstract |
This paper proposes a technique for automatic control of the bias voltage of a Mach-Zehnder modulator (MZM) using deep neural network-based loss function optimization. The proposed technique is applicable to both Intensity Modulation/Direct Detection (IM/DD) and Coherent systems. It extends traditional bias control techniques, which rely on optical power detection, by utilizing artificial intelligence (AI) to enhance modulation performance. To validate the proposed technique, the fundamental characteristics of MZMs and the operation of the Root Mean Square Propagation (RMSProp) loss function optimization technique are mathematically demonstrated. Simulation results related to automatic bias control are also presented, demonstrating the effectiveness of the technique. The key experimental results present the Bit Error Rate (BER) and Error Vector Magnitude (EVM) as functions of iterations. Even under continuous bias drift, the BER remains below 10-5, and the EVM around 10%, demonstrating the stability of the proposed technique. |