Frame domain signal processing: Framework and applications

Besides basis expansions, frames representations play a key role in signal processing. We thus consider the problem of frame domain signal processing, which is more complex and challenging than transform domain processing. Examples of such processing abound, from overlap-add/save convolution, to frequency domain LMS, and frame magnitude reconstruction. We develop a unified view of all these situations by using a common Hilbert space view of the problem, and consider algorithms in this common framework. In addition to a synthetic view of multiple signal processing methods in frames, we derive several original results. This include a direct solution to spectral modification (which usually uses an iterative algorithm) and a unicity condition for reconstruction from frame coefficient magnitudes.

[1]  J. Kovacevic,et al.  Life Beyond Bases: The Advent of Frames (Part II) , 2007, IEEE Signal Processing Magazine.

[2]  J.B. Allen,et al.  A unified approach to short-time Fourier analysis and synthesis , 1977, Proceedings of the IEEE.

[3]  Stephen A. Dyer,et al.  Digital signal processing , 2018, 8th International Multitopic Conference, 2004. Proceedings of INMIC 2004..

[4]  Jae S. Lim,et al.  Signal estimation from modified short-time Fourier transform , 1983, ICASSP.

[5]  J. Kovacevic,et al.  Life Beyond Bases: The Advent of Frames (Part I) , 2007, IEEE Signal Processing Magazine.

[6]  Roberto Cristi,et al.  On the Steady State Performance of Frequency Domain LMS Algorithms , 1993, IEEE Trans. Signal Process..

[7]  J. Shynk Frequency-domain and multirate adaptive filtering , 1992, IEEE Signal Processing Magazine.

[8]  Bernhard G. Bodmann,et al.  Fast algorithms for signal reconstruction without phase , 2007, SPIE Optical Engineering + Applications.

[9]  N. Aronszajn Theory of Reproducing Kernels. , 1950 .