Showing 1–2 of 2 results for author: Takumi, I

A Cutting Plane Method based on Redundant Rows for Improving Fractional Distance

Authors: Makoto Miwa, Tadashi Wadayama, Ichi Takumi

Abstract: In this paper, an idea of the cutting plane method is employed to improve the fractional distance of a given binary parity check matrix. The fractional distance is the minimum weight (with respect to l1-distance) of vertices of the fundamental polytope. The cutting polytope is defined based on redundant rows of the parity check matrix and it plays a key role to eliminate unnecessary fractional v… ▽ More In this paper, an idea of the cutting plane method is employed to improve the fractional distance of a given binary parity check matrix. The fractional distance is the minimum weight (with respect to l1-distance) of vertices of the fundamental polytope. The cutting polytope is defined based on redundant rows of the parity check matrix and it plays a key role to eliminate unnecessary fractional vertices in the fundamental polytope. We propose a greedy algorithm and its efficient implementation for improving the fractional distance based on the cutting plane method. △ Less

Submitted 17 July, 2008; originally announced July 2008.

Comments: 8 pages, To be presented at Turbo Coding 2008

Gradient Descent Bit Flipping Algorithms for Decoding LDPC Codes

Authors: Tadashi Wadayama, Keisuke Nakamura, Masayuki Yagita, Yuuki Funahashi, Shogo Usami, Ichi Takumi

Abstract: A novel class of bit-flipping (BF) algorithms for decoding low-density parity-check (LDPC) codes is presented. The proposed algorithms, which are called gradient descent bit flipping (GDBF) algorithms, can be regarded as simplified gradient descent algorithms. Based on gradient descent formulation, the proposed algorithms are naturally derived from a simple non-linear objective function. A novel class of bit-flipping (BF) algorithms for decoding low-density parity-check (LDPC) codes is presented. The proposed algorithms, which are called gradient descent bit flipping (GDBF) algorithms, can be regarded as simplified gradient descent algorithms. Based on gradient descent formulation, the proposed algorithms are naturally derived from a simple non-linear objective function. △ Less

Submitted 7 April, 2008; v1 submitted 2 November, 2007; originally announced November 2007.

Comments: 6 pages, 8 figures