ITR: Information-Theoretic Limits in Data Storage Systems

University Of California-San Diego, La Jolla CA

Investigators

Abstract

Data storage systems represent one of the pillars supporting the modern information age. Advances in computing and communication technologies, and the networks that incorporate them, have increased the demand for high capacity data storage devices in applications ranging from biological research (e.g., the visualization and analysis of massive multimedia data-sets) to consumer electronics (e.g., the storage of high-resolution image, audio, and video data in digital cameras). The research addresses analytic and numerical methods for determining the information-theoretic limits on achievable storage densities and data transfer rates in recording channels. It also develops and evaluates modulation, coding, and signal processing techniques that can approach or achieve these limits. The research considers both one-dimensional channel models applicable to "track-oriented" storage devices, such as disks and tapes, as well as higher-dimensional models relevant to exploratory page-oriented and volumetric storage technologies, such as holographic recording and thermomechanical recording based upon atomic force microscopy. More specifically, new information-theoretic and algorithmic methods are being used to study lower and upper bounds on the capacity of recording channels, using analytical and empirically-based models for both longitudinal and perpendicular recording systems. Bounds on the information rates of channels in two and higher dimensions are also being developed. In order to achieve performance approaching the theoretical limits, new signal processing and coding methods are required. In the area of constrained modulation coding, the project is exploring the analysis and design of one-dimensional constrained codes that more efficiently combine the functions of modulation and error-correction. The nascent theory of constrained coding in two and higher dimensions is also being investigated, with a focus on fundamental problems, such as representation of multi-dimensional constrained systems, determination of constraint capacity, and efficient encoding and decoding algorithms. Channel coding techniques based upon concatenation architectures, graphical code models, and iterative decoding have been shown to effectively achieve capacity on certain memoryless channels. The channel coding component of this project addresses theoretical and algorithmic problems associated with the application of these new and powerful coding concepts to data storage channels. In particular, the research considers the design and performance analysis of turbo-like and low-density parity-check codes for channels with intersymbol-interference, the trade-offs between latency and complexity in message-passing decoder architectures, and new approaches to combining coding, equalization, and detection that can more closely approach the recording channel capacity. The project also addresses fundamental problems of channel coding and detection in two-dimensional recording systems.