Finite word length analysis of the radix-22 FFT

In this paper, we analyze the quantization error effects of the radix-22 FFT algorithm. We propose per tone models for the error power. This is a different approach from the common choice of a maximum or mean value over the spectrum. In particular, we treat three different errors: due to input quantization, due to coefficient quantization and due to quantization after a multiplication. This analysis is applied to a DMT scheme. Simulation results agree with the theoretical predictions.

[1]  P. W. Wong.  Quantization noise, fixed-point multiplicative roundoff noise, and dithering , 1990, IEEE Trans. Acoust. Speech Signal Process..

[2]  John M. Cioffi,et al.  Understanding Digital Subscriber Line Technology , 1999 .

[3]  Alan V. Oppenheim,et al.  Discrete-Time Signal Pro-cessing , 1989 .

[4]  Mats Torkelson,et al.  A new approach to pipeline FFT processor , 1996, Proceedings of International Conference on Parallel Processing.

[5]  Shu Hung Leung,et al.  On the statistics of fixed-point roundoff error , 1985, IEEE Trans. Acoust. Speech Signal Process..

[6]  A. Oppenheim,et al.  Effects of finite register length in digital filtering and the fast Fourier transform , 1972 .