The fast-advancing modern nonvolatile memory technology includes flash memory and phase-change memory as the two most notable representatives. It is critical to create new schemes and architectures to improve their energy efficiency as well as their write speed, reliability and memory longevity. New schemes should fully utilize the unique properties of the memory technologies, including cell programming models, error models and endurance models. The key is to control cell states more flexibly and robustly by closely coupling novel coding schemes with the physics of these memory technologies.
In this project, a novel green storage technology will be developed based on coding theory for nonvolatile memory. The focus will be on flash memory and phase-change memory that share many common features. A new scheme for representing data using relative cell levels will be studied; it optimizes the energy efficiency for both writing and error correction. A novel variable-level cell technology will be developed to maximize storage capacity. New error-correcting codes will be designed to correct memory-specific errors. In addition, new processing techniques will be developed to balance the power consumption at peak and non-peak times.
The new designs will be integrated to form a comprehensive solution, and their performance will be analyzed thoroughly. The results will contribute to providing a new platform for high-performance ubiquitous computing based on nonvolatile memories.
