Dual-Input AWG with Adaptive Regression Algorithm for High-Precision FBG Interrogation

High-precision interrogation of fiber Bragg grating (FBG) sensors is critical for advanced sensing applications. We propose a time division multiplexing (TDM) interrogation scheme that combines a standard dual-input arrayed waveguide grating (AWG) with an adaptive LASSO-assisted polynomial regression algorithm, eliminating the traditional trade-off between hardware complexity and measurement accuracy. The core architecture leverages a high-speed optical switch to alternately excite two specifically positioned input ports of a standard AWG with a 1.6 nm channel spacing. Thus, it synthesizes an interleaved “virtual” channel grid with an effective 0.8-nm spacing without widening the optical passband. This relaxes the 3-dB bandwidth requirement for continuous interrogation by a factor of two and avoids the insertion loss penalty associated with passband broadening. Algorithmically, LASSO regularization enables self-adaptive selection of polynomial order while compensating for system nonlinearity and suppressing overfitting. Experiments demonstrate continuous interrogation over 34 nm with a root mean-square error (RMSE) of 0.31 pm and a wavelength resolution of 0.13 pm. Compared with single-input AWG schemes, the proposed TDM architecture simultaneously achieves higher accuracy, a wider dynamic range, and finer resolution without modifying the AWG layout, offering a practical route toward low cost and scalable high-precision fiber sensing networks.