In April 2007, the new Software Support for Metrology programme began. One of its themes is signal processing, specifically 'Quantitative approaches to digital signal processing in measurement systems'.

The project will develop rules and methods (including software tools) for the propagation of uncertainties through DSP-based measurement systems and for signal estimation (the recovery of signals from noisy or corrupted data and the removal of the effects of measurement systems on the underlying signal). It will provide an extensive set of signal processing tools that will be simple to use, reliable, and with quantifiable uncertainties, to ensure that best practice can be adopted easily by metrologists. The project will do this do this by concentrating on the development of:

• Design rules for DSP systems used in metrology, using simulations of measurement systems that rely substantially on digital signal capture and analysis. We will compare simulation results with analytical models where appropriate.
• Spectrum analysis tools and in particular the use of one- and two-dimensional Fourier-based methods in signal analysis, the derivation of transfer functions and the propagation of signals (including the angular plane wave spectrum method for predicting fields at locations other than the measurement plane).
• De-noising techniques using time-domain and frequency-domain methods applied to stationary and non-stationary signals. Wavelet methods may also be considered here, especially in the identification of weak non-stationary signals in noisy backgrounds.
• Deconvolution methods for calibration applications, including the derivation of transfer functions and the removal of instrumentation system responses from measurement data. Uncertainties in deconvolution processes are not yet understood, making their use in metrology currently problematic. This project aims to remedy this.

These scientific objectives will be delivered through a set of wide-ranging case studies in collaboration with metrologists from other National Measurement System programmes, National Measurement Institutes and industry that will demonstrate successful use of these tools in metrology applications.

• The first case study will apply simulation techniques, including Monte Carlo methods, to study the propagation of uncertainties through a complete DSP based measurement system with the aim of establishing design rules and uncertainty evaluation for DSP systems. In particular it will study the use of Fourier-based methods and deconvolution of instrument responses from measurements, with the aim of making deconvolution methods robust and reliable by introducing uncertainty evaluation methodologies that are suitable for non-expert users.
• The second case study will concentrate on traceable waveform capture and the associated DSP techniques. In particular it will examine traceability and uncertainty propagation in: demodulation systems, algorithms for noise reduction, spectrum analysis of non-stationary waveforms, and asynchronous sampling techniques.

As each case study is completed we will post relevant information in these web pages so that this signal processing resource will grow to reflect the signal processing work carried out at NPL.

If you would be interested in collaborating with us on any of the topics listed above, please contact us via this form