Interleaving and Error Concealment to Mitigate the Impact of Packet Loss in Resource-Constrained TDLAS/WMS Data Acquisition

Tomographic imaging of pollutant gas emissions from aeroengines is attractive for the development of engines and fuels. A 126-beam tomographic setup has previously been proposed utilizing tunable diode laser absorption spectroscopy aiming for fast spatially resolved measurement of CO 2 concentration. The custom data acquisition system uses a distributed architecture with at-site digital lock-in amplification, but remains resource constrained. A calibrated model is fitted to quadrature the first and second harmonic data, however, packet loss in ethernet and/or wireless networks can cause nondeterministic errors in the curve fitting and increased errors in recovered gas concentrations. Packet loss in this case, is a product of the available protocol, the high-vibration and high-noise industrial testing environment, the high network utilization expected, and the interrupt behavior of the embedded microprocessors. In this paper, the structure of the data acquisition system and the curve fitting approach are briefly discussed. Packet loss is then performed numerically to demonstrate the introduction of errors, as this cannot be swept experimentally without introducing other factors and increasing additive noise. An interleaving and error concealment mitigation approach is reported, that reduces this error, and can be applied to other resource-constrained remote acquisition systems such as Internet of Things applications. This approach is evaluated over parameters including extent of packet loss, interleaving ratio, and number of wavelength samples per packet. Viewing packet loss as a measurement SNR modifier, interleaving is shown to recover some SNR, but is ultimately limited. Processing of the received data using error concealment prior to spectrographic fitting is shown to increase tolerance.