2001 Fiscal Year Final Research Report Summary
RESEARCH ON VOLTERRA NONLINEAR ADAPTIVE SYSTEMS AND FAST ADAPTICE ALGORITHMS
| Project/Area Number |
12650395
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | CHUO UNIVERSITY |
|---|
Principal Investigator |
CHAO Jinhui Faculty of Science and Engineering, Professor, 理工学部, 教授 (60227345)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TSUJII Shigeo Faculty of Science and Engineering, Professor, 理工学部, 教授 (50020350)
|
|---|
| Project Period (FY) |
2000 – 2001
|
| Keywords | NONLINEAR SYSTEMS / ADAPTIVE FILTERS / SIGNAL PROCESSING / VOLTERRA SYSTEMS / FAST ALGORITHM / RLS ALGORITHM |
|---|
| Research Abstract |
It is shown that the error surface of quadratic Volterra adaptive filters (ADF) is always extremely steep in one particular direction but relatively flat in the rest directions. This nontrivial geometry explains the instability in the learning processing of Volterra ADF and implies and unavoidable slow convergence of tradictional gradient adaptive algorithms. On the other hand, the RLS algorithm for Volterra ADF costs O(N^5) multiplications where N is the number of linear terms in the filter input. This paper presents a new algorithm for Gaussian input signals which converges in the same rate as RLS but costs only O(N^2) multiplications, the same order as the LMS algorithm. This algorithm is based on a complete analysis of the intrinsic geometry of the error surface for white input signals and whitening operation of colored input signals. Simulations shown that this algorithm works well also in non-Gaussian input cases.
|
