Performance Characterization of LDPC Codes for Large-Volume NAND Flash Data

Abstract

High bit error rates of next-generation flash devices necessitate the use of more powerful error correction codes (ECCs), such as low-density parity-check (LDPC) codes, instead of the legacy Bose-Chaudhuri-Hocquenghem (BCH) codes. Unlike algebraic codes, the random nature of LDPC codes as well as their ability to use soft information requires the use of Monte Carlo (MC) simulations to evaluate code performance. Given a large volume of NAND data, this can pose resource challenges both in terms of simulation platforms and time needed for the Monte Carlo simulations. In order to overcome these challenges, we introduce a new channel metric in this paper to quantify the quality of soft information and propose a practical LDPC code performance characterization methodology.