A fast approximation for adaptive wavelength selection for tunable infrared chemical sensors

Abstract

Active mid-infrared (MIR) spectroscopy with tunable lasers is a leading technology for standoff detection of trace chemicals. Traditionally, the laser wavelength is swept to sample its tuning range uniformly without regard to how informative particular wavelengths may be. In contrast, this paper presents an adaptive technique to control the sequence of wavelengths, which the system samples in order to maximize chemical identification accuracy while minimizing measurement time and power consumption. Realtime adaptive wavelength selection is enabled by an approximation, which operates ∼40 000× faster than the full calculation. Application of this technique to synthetic data suggests that it can reduce the number of wavelengths required (and hence the measurement time) by a factor of two relative to an evenly spaced grid, with even higher gains for chemicals with weak signatures. While this paper focuses on MIR sensors using external cavity quantum cascade lasers, the technique is applicable to any system that has a tunable element and appropriate data characteristics.