AuthorKing, Brian Michael
AdvisorNeifeld, Mark A.
MetadataShow full item record
PublisherThe University of Arizona.
RightsCopyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
AbstractVolume holographic memories (VHMs) are a candidate technology for next-generation high-density and high data-rate digital storage. Capacities greater than 1 Terabit are promised, available at read-out rates exceeding 1 Gigabit per second. The capacity target will be achieved through two mechanisms. First, retrieval in a VHM reconstructs a holographic page (a two-dimensional image) captured on a CCD (charge-coupled device) camera. Each page represents on the order of one million bits of data, by encoding the data as bright and dark pixels in the 1024 x 1024 stored/retrieved image. Second, due to the thickness of the recording medium, a large number of such pages can be recorded in the same volume of material. In this dissertation we address some of the difficult technical issues that either currently limit the VHM system design, or are expected to become a limiting factor in the future. The first such concern involves how to process the simultaneous optical arrival of one million pixels. In high-density storage, there will be significant cross-talk between pixels which limits the storage capacity. We develop a novel highly-parallel focal-plane processor, which can significantly improve the system capacity by performing reliable detection in the presence of optical blur and alignment errors introduced by the imaging system. A fabricated proof-of-concept VLSI design is described. Another fundamental noise source is caused by the cross-talk between holographic pages. Reconstruction of the desired data page reconstructs every page in the memory, albeit at a very low relative diffraction efficiency. As the number of multiplexed pages increases, the cross-talk from the other pages can constitute a significant optical field noise source. Apodization seeks to either suppress this noise source or control it such that system tolerances can be relaxed. Bright data pixels are stored by altering the material properties of the crystal. However, dark pixels require no adjustment to the crystal; they are implicitly stored. This asymmetric storage cost drives a capacity improvement by biasing the data pages to contain more dark pixels and fewer bright pixels. An increased number of pages can be stored at the same reconstruction fidelity. We propose a novel modulation code to encode and decode these sparse data pages. Experimental results are presented showing the improvement in capacity. If the data page is composed of non-binary or grayscale pixels, then a further capacity enhancement is possible. The previous binary modulation code is extended into an arbitrary grayscale modulation code and a low-complexity maximum-likelihood decoder is developed as well as a mathematical proof of correctness. Extensive experimental results verify that the proposed method is practical and offers a substantial capacity improvement.
Degree ProgramGraduate College
Electrical and Computer Engineering