فارسی English
A new efficient fountain decoder for LT codes


دکتر یوسفی،
دانشگاه کوئینز کانادا

Iteratively-decoded low-density parity-check (LDPC) codes exhibit performance degradation manifested in the so-called ‘error floor’, corresponding to a significant flattening in the error probability curves. Although error-prone patterns have been extensively studied for low-density parity-check (LDPC) codes, they have never been fully explored for low-density generator matrix (LDGM) codes. It is shown that these codes suffer significantly at low error rates from certain combinatorial structures within the tanner graph representation of the code.
In this work, we discuss one important combinatorial object referred to as an absorbing set (AS) in the Tanner/factor graph representation of an LDGM code. We then demonstrate how these absorbing sets are damaging to the realized rate, delay, and decoding cost of fountain LDGM codes. This culminates in our main results including a new stopping criteria for the iterative fountain decoders as well as a new decoder. We observe 20% reduction of the decoding complexity for a Luby-Transform (LT) code while at the same time achieving coding gains of 0.6 dB and 0.9 dB at bit error rates of 10-5 and 10-6, respectively, when compared against the traditional decoders.










بیستمین کنفرانس مهندسی برق ایران

ICEE 2012، تهران، ایران