A probability density estimation algorithm on multiwavelet for the high-resolution ADC

Abstract

ADC (Analog-to-digital converter) is an important component in measuring instruments that converts the analog signals to digital form. The actual effective vertical resolution of ADC is one of the most important indicators for evaluating the ADC performance. However, it is restricted by various factors such as external disturbance, environmental interference, clock jitter, transmission delay, etc. These problems lead to a lower ADCs’ effective resolution than the ideal value. This paper proposes an improved algorithm dedicated to decrease errors in measured signals. In this algorithm, we make Multiwavelet decomposition to get the probability density function of measured signal first, then use singular value decomposition to remove the noise related frequency characteristic values in the multiwavelet basis coefficients, next create a Bayesian fusion iteration process to optimize the probability density function and build a confidence interval for elimination of the noise data, and finally the signal is reconstructed and average filtered for output. Through simulation experiments we show an improvement of the vertical resolution and SNR (signal-to-noise ratio) in comparison to the conventional averaging filter, high resolution acquisition and oversampling methods.