1998 Fiscal Year Final Research Report Summary
Parallel Module Implementation of Filter Banks and Its Applications
| Project/Area Number |
09650431
|
|---|
| 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 | Kanazawa Institute of Technology |
|---|
Principal Investigator |
MURAKAMI Hideo Kanazawa Institute of Technology Dept.of Engineering, Professor, 工学部, 教授 (90113034)
|
|---|
| Project Period (FY) |
1997 – 1998
|
| Keywords | Filter bank / Fast algorithm / Signal analysis / Multiplexing system |
|---|
| Research Abstract |
A filter bank consists of an analysis filter bank which divides an input signal into a number of subband signals carrying prescribed frequency band energies of the input, and a synthesis filter bank which reconstructs the input from the subband signals. A filter bank which reconstruct the original signal with no distortion when the synthesis filter bank is directly connected to the outputs of the analysis filter bank is called a perfect reconstruction (PR) filter bank. The PR condition is required for to be practical in many applications, including communication and subband data compression applications. This research project has achieved meaningful contribution to the implementation and PR condition problems of filter banks. The filter bank is decomposed into a set of module filter banks of small size in way such that available fast processors can be incorporated in the implementation. This property is particularly advantageous for fast parallel hardware realization. A careful investigation shows the propose d implementation scheme renders much less restrictive PR condition. This eased restriction makes possible to design better filter frequency responses for a fixed filter length. Followings are specific results.
(1) A new polynomial orthogonal expansion for a digital signal is derived.
(2) Fast algorithms are derived based on the new polynomial expansion.
(3) Parallel module scheme is established for implementation a maximally decimated filter bank.
(4) PR condition of the proposed implementation is analyzed, and the condition is shown to be less-restrictive than a conventional filter bank
(5) The obtained implementation is applied for multiplexing communication system.
|
