9406197 O'Sullivan The existence, at Washington University, of strong programs both in information theory and communications systems, and in magnetic information storage, has resulted in fruitful collaboration on the information capacity of magnetic storage systems. The initial results of this collaboration show that the ultimate capacity bounds of magnetic storage media exceed existing practice and short term projections by nearly three orders of magnitude. The capacity bounds of the complete storage system reflect these ultimate bounds, but are further limited by the other system components, including transducers, codes, and electronics. This project analyzes these limits, and seeks design rules for components and channels to maximize the use of the potential capacity. This objective is supported by three connected research activities. The first of these activities addresses the modeling of storage media. The models used to date have been idealized to reduce their computational complexity; they have yielded valuable qualitative results. The project is developing a quantitative analysis requiring more realistic modeling while still being computationally tractable. Specifically, thermal, spatial and dynamic effects will be included in more refined medium and channel models. Annealing algorithms will be incorporated to examine this area. This will allow understanding of the fundamental limits of this channel. The analysis of possible implementations of a complete storage channel from the viewpoint of its information capacity comprises a major part of our research. Such an analysis accounts not only for the properties of the medium, but also for such issues as asymmetries of storage and retrieval, electronics, and archivability. Initial studies focus on computing the capacity of simplified versions of the recording channel that include the read head. A striking result of this work has been the development of relations and equivalences between the ph ysical parameters of storage media and their capacity bounds. This has led to a proposal of new medium design guides for ultra-high density storage. These results were developed for media imbedded in ideal systems, and are expected to be reflected in similar relations in existing and projected technological realizations. The equivalences will be tested experimentally, using a variety of media, in their unique laboratory facilities. These researchers were the first to demonstrate experimentally that medium noise has a repeatable component due to local medium properties. These experiments motivated the development of medium models for use in magnetic recording system analysis and design, which led to the proposal to exploit that deterministic noise component to increase the capacity of magnetic recording systems. Experiments form an integral part of the research. The research team has theorists working closely with experimentalists. As a result, the models and results derived can immediately be compared to measurements of physical systems. The supporting studies will combine to address the design of optimal signal processing schemes, for realizable systems, that take full advantage of the available components. The design strategies will explore new methods for storing information magnetically, and unconventional information processing techniques. It is hoped that the results will have wide applicability to other information systems; the approach, joining device physics with information science, is unparalleled in the field. This physically-based approach to information science will develop new analysis techniques for complex systems, motivate the design of new components and systems, and design new signal processing strategies to approach the computed capacity limits. Through this novel and unconstrained approach new ways to process and store information will emerge yielding dramatic increases in storage density. ***
