1995 Fiscal Year Final Research Report Summary
RESEARCH ON A SOFT-DECISION ITERATIVE DECODING OF A LINEAR BLOCK CODE
| Project/Area Number |
06650416
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | Nara Institute of Science and Technology |
|---|
Principal Investigator |
KASAMI Tadao Nara Institute of Science and Technology Graduate School of Information Science Professor, 情報科学研究科, 教授 (50029378)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
FUJIWARA Toru Osaka University Faculty of Engineering Science Associate Professor, 基礎工学部, 助教授 (70190098)
TAKATA Toyoo Nara Institute of Science and Technology Graduate School of Information Science, 情報科学研究科, 助教授 (50216652)
YAMAMOTO Heiichi Nara Institute of Science and Technology Graduate School of Information Science, 情報科学研究科, 教授 (40243357)
|
|---|
| Project Period (FY) |
1994 – 1995
|
| Keywords | linear block codes / suboptimum decoding / soft-decision / decomposable codes / iterative decoding / decoding complexity |
|---|
| Research Abstract |
In this research, we investigate the following two types of soft-decision suboptimum decoding of a linear block code.
(a) Generalized closest coset decoding for a decomposable code : For a code C of length n, C is said to be decomposable if C is the direct sum of C_1 and C_2 of length n. For this type of decomposable codes, we propose a new suboptimum closest coset decoding scheme shown below. Let C'_2 be a supercode of C_2 which is independent from C_1. The following is a kind of closest coset decoding of C.Let z be the received unquantized n-tuple. (1) Decode z into a codeword in C_1 + C'_2 by maximum likelihood decoding. Let u + v be the decoded codeword in C_1 + C'_2, where u*C_1 and V*C'_2. (2) Decode z into a codeword in C_2 + {u}, a translate of C_2, by maximum likelihood decoding. Let u+v denote the decoded codeword, which is the output of the overall decoding. This suboptimum decoding scheme is a generalization of the multi-stage decoding scheme for multi-level modulation codes
… More
and the closest coset decoding scheme for codes with the |u|u+v|-structure proposed by Hemmati.
We investigate the evaluation method of the block error probability of a decomposable block code for the above decoding scheme over an AWGN channel. For some specific example codes, we evaluate the block error probability and the decoding complexity.
(b) Iterative decoding : Consider the following iterative decoding scheme : First generate an initial candidate codeword by a relatively less complex decoding a algorithm, such as a hard-decision algebraic decoding. Then iterate that we find more likely codeword around the current candidate codeword by searching by using a minimum weight subtrellis diagram around the current candidate codeword, until we can't find more likely codeword, neither the candidate codeword pass a test of optimality.
For this type of iterative decoding, we derive the least stringent sufficient condition that the available information on the code is restricted to (i) the minimum weight and a few small weights and (ii) for a given positive integer h, h or fewer already generated candidate codewords. We also derive a sufficient condition on the optimality under the same assumption as above, except that the condition (i) is changed to (i') complete knowledge of the weight profile of a code is known. We evaluate the average reduction of number of decoding iterations for some Reed-Muller codes over an AWGN channel.
We also present a new soft-decision iterative decoding algorithm based on searching by using a minimum weight subtrellis diagram. We show the block error probablity and the decoding complexity for some Reed-Muller codes over an AWGN channel for the proposed decoding. Less
|
