Abstract | ||
---|---|---|
We present two algorithms for fast time-domain Volterra filtering. The first algorithm computes the required products of input samples using only one multiplication per term. Since the products are explicitly computed, this algorithm can be used for adaptation as well as for filtering. The second algorithm generalizes Horner's method for polynomial evaluation and directly computes output samples without computing products of input samples. This way, time-domain Volterra filtering can be implemented with only one multiplication per parameter. We implemented several variants of the proposed algorithms in C and compared their runtime, demonstrating a speedup up to 5. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1109/ACSSC.2016.7869029 | 2016 50th Asilomar Conference on Signals, Systems and Computers |
Keywords | Field | DocType |
fast time-domain filtering,Volterra filtering,multiplication per term,generalize Horner method,polynomial evaluation | Frequency domain,Time domain,Kernel (linear algebra),Mathematical optimization,Polynomial,Computer science,Filter (signal processing),Algorithm,Multiplication,Speedup,Computational complexity theory | Conference |
ISSN | ISBN | Citations |
1058-6393 | 978-1-5386-3955-9 | 1 |
PageRank | References | Authors |
0.45 | 8 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
H. Enzinger | 1 | 13 | 3.29 |
Karl Freiberger | 2 | 2 | 0.83 |
Gernot Kubin | 3 | 197 | 38.65 |
Christian Vogel | 4 | 361 | 28.88 |