Reconstruction Of Non-Uniformly Sampled Signals Using Gerchberg-Papoulis Method

Abstract

Analysis of non-uniformly sampled signals is often severely limited, since most signal processing methods rely on constant sampling period. If we still want to apply these methods, the signal must be resampled. Gerchberg-Papoulis algorithm is a method of signal reconstruction. It is commonly used for band-limited extrapolation of uniformly sampled data. We show that it is suitable for reconstruction of non-uniformly sampled signals as well. Our target application is reconstruction of time series measured by a car driving simulator. To demonstrate the benefits of band-limited reconstruction, we compare it with standard interpolation methods. The main advantage of the proposed algorithm is its ability to deal with noise and sampling jitter.