Evaluation of the Low Frame Error Rate Performance of LDPC Codes Using Importance Sampling
Dolecek, L.
Zhengya Zhang
Wainwright, M.
Anantharam, V.
Nikolic, B.
Univ. of California, Berkeley;
This paper appears in: Information Theory Workshop, 2007. ITW '07. IEEE
Publication Date: 2-6 Sept. 2007
On page(s): 202-207
Location: Tahoe City, CA,
ISBN: 1-4244-1564-0
INSPEC Accession Number: 9827068
Digital Object Identifier: 10.1109/ITW.2007.4313074
Current Version Published: 2007-09-24
Abstract
We present an importance sampling method for the evaluation of the low frame error rate (FER) performance of LDPC codes under iterative decoding. It relies on a combinatorial characterization of absorbing sets, which are the dominant cause of decoder failure in the low FER region. The biased density in the importance sampling scheme is a mean-shifted version of the original Gaussian density, which is suitably centered between a codeword and a dominant absorbing set. This choice of biased density yields an unbiased estimator for the FER with a variance lower by several orders of magnitude than the standard Monte Carlo estimator. Using this importance sampling scheme in software, we obtain good agreement with the experimental results obtained from a fast hardware emulator of the decoder.
Index
Terms
Available to subscribers and IEEE members.
References
Available to subscribers and IEEE members.
Citing Documents
Available to subscribers and IEEE members.
Cart
