Abstract
This paper presents two types of protograph-based globally-coupled low-density parity-check (GC-LDPC) codes formed by a new edge spreading operation. This operation is called the global edge spreading. The Gaussian approximation (GA) and the protograph-based extrinsic information transfer (P-EXIT) analysis are then generalized over a special type of burst-erasure channels (BuECs). Such channel incorporates both Gaussian noise and burst erasures, and is denoted by the Gaussian channel with burst erasures (BuEC-G). Furthermore, the stability condition for BuECs-G is proved and an edge spreading optimization method is proposed to design the structured GC-LDPC codes by predicting the iterative decoding thresholds of corresponding protographs. Simulation results show that the optimized GC-LDPC codes can achieve better thresholds and error performances than existing well-designed GC-LDPC codes, and provide near-capacity performances over BuECs-G.
Highlights
For the data transmission, data packets are inevitablely influenced by both random noise and interference
We focus on a special burst-erasure channel (BuEC) which incorporates both random noise (Gaussian noise) and erasures
Based on the Gaussian approximation (GA) and P-extrinsic information transfer (EXIT) analysis, we present an edge spreading optimization method for minimizing the gap between the capacity and the iterative decoding threshold of globally-coupled low-density parity-check (GC-low-density parity check (LDPC)) codes for a given range of code rates and code lengths over BuECs-G
Summary
Data packets are inevitablely influenced by both random noise and interference (burst-noise). Taking the magnetic (or optical) recording system as an example, we can regard its background noise as the white Gaussian noise, and designate the detected thermal asperities (or scratches) at the decoder as erasures [4], [5]. The noise in such channel is the combination of the background Gaussian noise and erasures, this is different from the classic erasure-burst channel (defined in [1]) that only considers erasures. We refer to this type of channels as the Gaussian channel
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have