Title :
An Analysis of the Sum-Product Decoding of Analog Compound Codes
Author :
Fangning Hu ; Henkel, W.
Author_Institution :
Int. Univ. of Bremen, Bremen
Abstract :
We investigate the sum-product decoding on graphs of analog compound codes and show that the iterative decoding can be completely analyzed by tracing the mean vector at each iteration. A novel geometric analysis is proposed to visualize the iterative decoding process in the Euclidean space. Based on this geometric analysis, we propose to decompose the analog compound codes into several orthogonal constituent code spaces to achieve the fastest convergence speed. Simulations are given to verify our conclusions.
Keywords :
geometry; graph theory; iterative decoding; parity check codes; turbo codes; Euclidean space; LDPC codes; analog compound codes; cycle-free graph; geometric analysis; iterative decoding; orthogonal constituent code space; sum-product decoding; turbo codes; AWGN channels; Algorithm design and analysis; Block codes; Convergence; Iterative algorithms; Iterative decoding; Jacobian matrices; Parity check codes; Springs; Sum product algorithm;
Conference_Titel :
Information Theory, 2007. ISIT 2007. IEEE International Symposium on
Conference_Location :
Nice
Print_ISBN :
978-1-4244-1397-3
DOI :
10.1109/ISIT.2007.4557496
