Abstract
Dropouts can greatly affect the accuracy of flow measurements in the case of low-SNR Doppler signals. The dependence of the measurement accuracy on easily determined dropout parameters has been studied theoretically and experimentally. Based on a Gaussian narrowband representation of the Doppler signal, the dropout process is characterised by two parameters, tracking rate and unlock frequency and modelled as alternating renewal process. It is found that velocity estimates are poorest for unlock frequencies of the order of the signal bandwidth but improve at higher and lower values. Laminar flow experiments indicate that the transitions between lock and unlock are best described by a Poisson process. An application of this finding to the determination of the LDV scattering volume demonstrates that even under adverse tracking conditions useful volume-dimension estimates may be obtained.