Linear and Cyclic Convolutions
Linear convolution is one of the most frequent computations carried out in digital signal processing (DSP). The standard method for computing a linear convolution is to zero-tap, turning the linear convolution into a cyclic convolution, and to use the convolution theorem, which replaces the cyclic convolution by an FT of the corresponding size. In the last ten years, theoretically better convolution algorithms have been developed. The Winograd Small Convolution algorithm  is the most efficient as measured by the number of multiplications.
KeywordsDigital Signal Processing Polynomial Ring Toeplitz Matrix Circulant Matrix Convolution Theorem
Unable to display preview. Download preview PDF.
- Auslander, L., Cooley, J. W. and Silberger, A. J. “Number Stability of Fast Convolution Algorithms for Digital Filtering”, in VLSI Signal Proc., IEEE Press, 1984, pp. 172-213.Google Scholar
- Blahut, R. E. Fast Algorithms for Digital Signal Processing, Addison-Wesley, 1985, Chapters 3 and 7.Google Scholar
- Nussbaumer, H. J. Fast Fourier Transform and Convolution Algorithms, Second Edition, Springer-Verlag, 1981, Chapters 3 and 6.Google Scholar
- Burrus, C. S. and Parks, T. W. DFT/FFT and Convolution Algorithms, John Wiley and Sons, 1985.Google Scholar
- Oppenheim, A. V. and Schafer, R. W. Digital Signal Processing, Prentice-Hall, 1975.Google Scholar