Computer science's Jiang wins IEEE best paper award

Anxiao (Andrew) JiangAssistant Professor Anxiao (Andrew) Jiang is a recipient of the 2009 IEEE Best Paper Award in Signal Processing and Coding for Data Storage. The award is sponsored by the Data Storage Technical Committee (DSTC) of the IEEE Communications Society. Every year, one best paper in the area of data storage is selected for this prestigious award.The prize-winning paper, "Rank Modulation for Flash Memories," was authored by Jiang, Robert Mateescu, Moshe Schwartz and Jehoshua Bruck. It appeared in the IEEE Transactions on Information Theory, June 2009. In the paper, a new data representation paradigm was presented that uses the relative order of cell levels to store data. It is very distinct from the current flash memory technologies, where fixed cell levels are used in single-level cells (SLC) and multi-level cells (MLC) to represent data. The rank modulation scheme can make writing faster and more robust in flash memories.The basic storage element of a flash memory is a floating-gate cell. Electrons can be injected into it and remain there. The number of electrons in a cell determines its threshold voltage, which can be mapped to data and be measured. The cells in a flash memory are partitioned into blocks, where many (e.g.,  or more) cells form a block."Flash memories have a special feature called block erasure," says Jiang. "It means that to remove electrons from any cell, the electrons in the whole block have to be removed together. This makes data storage, especially writing, very challenging."For instance, when writing data into a block, if too many electrons are injected into even one cell, the whole block has to be erased and reprogrammed. To minimize the chance of over-injection, cells have to be programmed slowly using multiple rounds of injection. But this approach is becoming increasingly costly when more bits are stored in a cell for higher capacity.This paper proposes a different method to store data. By mapping data to the relative order of cell levels, writing becomes flexible and free of over-injection errors. The rank modulation scheme also leads to new solutions for data modification and error correction. They reveal new directions for next-generation flash memories, which have a vast market.Anxiao (Andrew) Jiang is an Assistant Professor of Computer Science and Engineering at Texas A&M University. He received his Ph.D. in Electrical Engineering from California Institute of Technology in 2004. He joined Texas A&M University in 2005. His research interests include information theory, data storage in memories, networked systems, and algorithms.