Endurance-Limited Memories with Informed Decoder

Abstract

Non-volatile resistive memories, such as phase change memories and resistive random access memories, have attracted significant attention recently due to their scalability, speed, and rewritability. However, in order to use these memories in large-scale memory and storage systems, the limited endurance deficiency of these memories must be addressed. In a recent paper, we proposed a new coding scheme, called endurance-limited memories (ELM) codes, which increases the endurance of these memories by limiting the number of cell programming operations. Namely, an l-change t-write ELM code is a coding scheme that allows to write t messages into some n binary cells while guaranteeing that the number of times each cell is programmed is at most l. There are several models of these codes which depend upon the information that is available to the encoder and the decoder before each write. This information can be one of the following three options: 1. the number of times each cell has been programmed, 2. only the memory state before programming, or 3. no information is available on the cells’ state or previous writes. In this paper, we study the models in which the decoder knows on each write the number of times each cell has been programmed before the last write, while for the encoder we consider the aforementioned three possibilities.