|
|
|
“Three-Dimensional Detection and Equilization for Page-Oriented Optical Data Storage Systems”
by Tawei Ho
August 2005
In this thesis, we describe the effects of both inter-page interference (IPI) and inter-symbol interference (ISI) in volumetric (3D) page-oriented optical data storage (PODS) systems, and present a detailed mathematical model for 3D PODS systems. Based on this model, we describe the equalization and detection techniques that overcome both IPI and ISI in the readout system. This detection scheme is extremely useful for future high data capacity page-oriented storage systems, where the IPI effect is severe. For different application requirements, our proposed methods can be operated flexibly in both batch and recursive modes. In our simulation results, our detection method shows consistent improvements over conventional methods.In addition, we propose multi-level data signaling on 3D PODS to enhance the data density further. Our detection algorithm also yields significant improvements on multi-level 3D PODS systems. Therefore, it is advantageous to apply our detection method for increasing the storage density in future storage systems. We also propose a variable thresholding detection method with improved performance. It overcomes the ISI and IPI effects simultaneously by using knowledge of the interference from surrounding neighbors and changing the threshold locally. Hence, it provides a low complexity detection method for the readout system.To further reduce the complexity, we examine the implementation of least square equalization and variable threshold methods using finite precision arithmetic. In a bit-error-rate comparison, we show that the fixed-point variable threshold method achieves similar results as the floating-point version but only using one seventh of the bits. Therefore, a potential saving of the storage and complexity is achieved by using finite precision arithmetic. Last, we investigate 3D PODS systems with pixel misregistration. Using our least square equalization and variable threshold methods, we have significant improvement when the amount of misalignment is 10% of the pixel pitch or less. These simulation results show the robustness and feasibility of our detection methods.