LAUSR

An optical vector analyzer (OVA) based on orthogonal polarization interrogation and polarization diversity detection is widely used to measure an optical device's loss, delay, or polarization-dependent features. Polarization misalignment is the OVA's primary error source. Conventional offline polarization alignment using a calibrator greatly reduces the measurement reliability and efficiency. In this Letter, we propose an online polarization error suppression method using Bayesian optimization. Our measurement results are verified by a commercial OVA instrument that uses the offline alignment method. The OVA featuring online error suppression will be widely used in the production of optical devices, not just in the laboratory.