Joint Estimation of Multiple Time Interleaved ADC Timing Offsets Based on Fourier Series Decomposition

Abstract

A method for reducing the complexity of estimating multiple timing offset in time interleaved analog to digital converter based on Fourier series decomposition is proposed. The timing offset of different slices of the time interleaved analog to digital converter can be understood as points on a curve, which represents the timing offset per slice. Thus, the estimation of the timing offset can also be achieved by fitting of the timing-offset-curve. The timing offset curve can be represented by a kernel regression coefficient which are orthonormal on a limited set, thus replacing the estimation of timing offset per slice with the estimation of the regression kernel coefficients. In this paper, the idea of estimating kernel coefficients by using Fourier series analysis is presented and analyzed. When the timing offset of different slices are highly correlated, it is shown that an improvement of 3 dB can be achieved with ~75% reduction in complexity, 6 dB improvement with ~60% reduction and 10 dB with ~40% reduction in respect to similar linear methods.