A Multidimensional Network Framework for Trapping Set Analysis

Talk by Allison N. Beemer

Speaker invited by: Prof. Dr. Joachim Rosenthal

Date: 29.03.17  Time: 16.00 - 17.00  Room: Y27H28

Persistent error floors in the Bit Error Rate curves of low-density parity-check codes with message-passing decoders have been shown to be characterized by graphical (sub)structures in the codes' Tanner graphs, such as pseudocodewords, absorbing sets, stopping sets, and trapping sets. In this talk, I will present a multidimensional network model as a means to analyze decoder failure and characterize trapping sets of graph-based codes. I will then illustrate this framework with applications to several code constructions, including product, half-product, and protograph codes.