Fiber sensing using wavelength-swept lasers: A compressed sampling approach

A compressive sampling (CS) technique for wavelength-swept fiber sensing is presented. Fiber sensors embedded into composite materials are key elements in smart structures and wavelength-swept lasers provide a favorably high local SNR. The vast amount of data produced by these systems can be handled by CS-based acquisition while sparse processing techniques provide high-resolution capabilities. We identify an appropriate overcomplete dictionary for wavelength-swept fiber sensing and adopt an adaptive OMP algorithm to alleviate the problem of coherent dictionary atoms. The presented technique incorporates an iterative method to estimate uncertain model parameters based on minimizing the residual signal. It features superior performance as compared to the case where a simplified Gaussian model is used and works successfully for simulated and experimental data at moderate SNRs.