Coding and signal processing has been serving as a cost-effective means to mitigate various types of noise and interferences, furthermore ensuring fast and reliable data recovery in data storage systems. For NVM, due to its specific characteristics and potential applications, in addition to the requirement of a high recording density, a large endurance and retention time, fast read/write access time, as well as high power efficiency are also expected. The NVM coding and signal processing team is working on innovative and efficient channel algorithms and techniques to meet all the above requirements. Our major research focus is as follows.
To model the major characteristics and anticipated impairments of PCRAM and STT-STT-MRAM channels
The modeling of PCRAM and STT-MRAM devices and systems is the prerequisite for the development of channel algorithms. Up till now, there is no channel model available for the PCRAM/STT-MRAM signals. Therefore, working closely with DSI’s PCRAM and STT-MRAM teams, the modeling team will develop appropriate channel models which include the major characteristics and anticipated impairments of PCRAM and STT-MRAM devices and systems. The channel modeling will be carried out by using both empirical data based approach and memory-physics based approach. It will serve as the basis for the development of codes and detectors for NVM.
To design and optimize various channel coding schemes and detection approaches to stretch the maximum possibilities of PCRAM and STT-MRAM
Based on the proposed PCRAM and STT-MRAM channel models, various channel algorithms and techniques will be developed accordingly. This includes hard error correction codes (ECCs), low-density parity-check (LDPC) codes, storage codes, as well as soft-output channel detectors. Our work will help to stretch the maximum possibilities of PCRAM and STT-MRAM, from a coding and signal processing perspective.
Fig. 1 Coding and signal processing for PCRAM/STT-MRAM channels
To provide valuable guidelines to the PCRAM and STT-MRAM device and system design
Advance coding and signal processing techniques may improve the system’s performance, increase the recording density, and extend the lifetime of the NVM. This may relax the engineering requirements on the cell’s expected physical behavior. By providing the ultimate performance measurements of PCRAM and STT-MRAM systems in terms of bit error rates and page error rates, this research will help the NVM device and system designers to evaluate and verify their designs.