Multidimensional IIR filters and robust rational interpolation
It is well-known that IIR filters can have a much lower order than FIR filters with the same performance. On the downside is that the implementation of an IIR filter is an iterative procedure while that of an FIR filter is a one-shot computation. But in higher dimensions IIR filters are definitely more attractive. We offer a technique where the filter’s performance specifications, stability constraints, its convergence speed and a protection against possible adverse effects of perturbations are all included in the design from the start. The technique only needs an off-the-shelf LP solver because the filter is obtained as a Chebyshev center of a convex polytope. The method deals with general non-causal non-separable filters.
KeywordsMultidimensional digital filters Non-causal IIR Linear programming Quadratic programming Rational approximation
Unable to display preview. Download preview PDF.
- Cuyt, A. & Salazar Celis, O. (2012). Multivariate data fitting with error control. TR Universiteit Antwerpen (in preparation).Google Scholar
- Harris D. (1981) Design of 2-D rational digital filters. Acoustics, Speech, and Signal Processing 6: 696–699Google Scholar
- Lu, W. S., & Hinamoto, T. (2011). Minimax design of stable IIR filters with sparse coefficients. In: International Symposium on Circuits and Systems (ISCAS 2011), May 15–19 2011. Brazil: Rio de Janeiro, pp. 398–401. IEEE. http://dx.doi.org/10.1109/ISCAS.2011.5937586 .
- Piyachaiyakul, N., & Charoenlarpnopparut, C. (2011). Design of three-dimensional adaptive IIR notch filters. Multidimensional Systems and Signal Processing, 1–12. doi:10.1007/s11045-011-0170-3 .