Reconstruction of Nonuniformly Sampled Signals

Project Type: Master/Diploma Thesis
Student: Tertinek Stefan

 

 In digital signal processing, the standard method to convert an analog signal into a digital signal is uniform sampling; it refers to the process of taking samples at uniformly spaced time instances. In many practical applications, however, sampling occurs at nonuniformly spaced time instances, and the goal is to reconstruct the original analog signal from the nonuniform samples. In this work, we consider the case when the nonuniform sampling instances slightly deviate from the uniform sampling instances and are known. We propose the differentiator-multiplier cascade, a novel reconstruction system which recovers the uniform samples from the nonuniform samples. The system consists of several stages where in each stage blocks of finite impulse response (FIR) filters designed as differentiators and time-varying multipliers improve the reconstruction. Contrary to the methods proposed so far, the main advantage of the reconstruction system is that it is not limited to the case of periodic nonuniform sampling. Furthermore, once the differentiators have been designed, they can be implemented with fixed multipliers, and only the coefficients of the time-varying multipliers have to be adapted when the sampling pattern changes; this reduces implementation costs substantially, especially in a practical application such as time-interleaved analog-to-digital converters (TI-ADCs).