Joint Decoding of Content-Replicated Codes on AWGNC for Flash Memory

Abstract

Continuously increasing capacity and scales have made flash memory an affordable product for both consumer electronic and system storage. However, as flash density increases, flash memory becomes more subject to noise which leads to the degrading of data reliability and endurance. Error-correcting codes and memory scrubbing are two approaches to handle data reliability issues. Specifically, in flash memory, the “out-of-place” updating mechanism leads to the existence of content-replicated error-correcting codes (ECC) that contain the same data. The thesis proposes a joint-decoding scheme using those content-replicated ECCs to further enhance the reliability of data in flash memory. Three different categories of content-replicated ECCs are explored and analyzed. The density evolution analysis results show that using content-replicated ECCs with the corresponding joint-decoding algorithms can effectively improve the error-correcting performance. Additionally, it is shown that increasing the diversity of content-replicated ECCs with some limits may further extend the error-correcting ability.