Signal/Noise Modeling and Detector Design for Media Noise Dominated Magnetic Recording Systems

介质噪声主导的磁记录系统的信号/噪声建模和检测器设计

• 批准号: 9904458
• 负责人: Aleksandar Kavcic
• 金额: $ 34万
• 依托单位: Harvard University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1999
• 资助国家: 美国
• 起止时间: 1999-08-15 至 2003-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9904458&HistoricalAwards=false
• 关键词: Signal; Noise; Modeling; Detector; Design

The object of the project is the development of the necessary tools and know-how to design and implementable signal detector for magnetic recording systems operating in signal-dependent correlated noise environments, i.e., media noise dominated environments. According to recent studies, it is anticipated that the current magnetic recording technology will hit the superparamagnetic limit in approximately five years. The superparamagnetic limit occurs when the magnetic media particles become so small that thermal energy at room temperature is enough to destabilize the particle by randomly switching its magnetization orientation. To push back the approaching superparamagnetic limit, signal processing circuits will have to cope with a radically changed signal/noise environment: the signal-to-noise ratio will drop, the magnetic media noise will dominate over the electronics white noise, and the overall noise will become extremely correlated and signal-dependent due to the intersymbol interactions in the magnetic media and due to the coloring introduced by the read head. This project will develop a fundamental understanding of nonstationary stochastic modeling and use it to develop a magnetic recording detector ideally tuned to correlated nonstationary media noise.The project will be conducted through collaboration between Harvard University and IBM Almaden Research Center. IBM, a recognized global magnetic recording technology leader, is in a unique position to provide the cutting edge experimental support required establishing the link between the theoretical and experimental parts of the project. The project combines three major efforts. (i) A understanding of signal-dependent noise modeling issues will be developed. This is an extension of widely used autoregressive modeling methods in stationary processes to nonstationary processes. Particular attention will be given to objective model-size estimation through the adaptation of Akaike's information theoretic criterion or Rissanen's minimum description length principle. (ii) The receiver will be designed through a complexity-constrained optimization procedure. Since both the intersymbol interference length and the noise memory length contribute equally to the receiver complexity, the constrained-complexity detector will minimize the sum of the two lengths under a performance specification. (iii) The final goal is the implementation of the detector in VLSI technology. Towards this goal, a software detector and then a hardware detector will be developed. Both detectors will be tested in real channel environments at IBM Almaden Research Center.

该项目的目标是开发必要的工具和专门知识，以设计和实现在信号相关噪声环境中运行的磁记录系统的信号检测器，即，媒体噪声主导环境。 根据最近的研究，预计目前的磁记录技术将在大约五年内达到超顺磁极限。 当磁性介质颗粒变得如此之小，以至于室温下的热能足以通过随机切换其磁化取向而使颗粒不稳定时，发生超顺磁性极限。 为了推回接近的超顺磁极限，信号处理电路将不得不科普急剧变化的信号/噪声环境：信噪比将下降，磁介质噪声将主导电子白色噪声，并且由于磁介质中的符号间相互作用以及由于读取头引入的着色，整体噪声将变得极其相关和信号依赖。 本项目将发展对非平稳随机建模的基本理解，并利用它开发一种磁记录检测器，该检测器理想地调谐到相关的非平稳介质噪声。该项目将通过哈佛大学和IBM Almaden研究中心之间的合作进行。 IBM是公认的全球磁记录技术领导者，在提供建立该项目的理论和实验部分之间的联系所需的尖端实验支持方面处于独特的地位。 该项目结合了三大努力。(i)将发展对信号相关噪声建模问题的理解。 这是在平稳过程中广泛使用的自回归建模方法到非平稳过程的扩展。 特别注意将给予客观的模型大小估计，通过适应赤池的信息理论标准或Rissanen的最小描述长度原则。 (ii)接收器将通过复杂度约束优化过程进行设计。 由于符号间干扰长度和噪声记忆长度对接收机复杂度的贡献相等，因此约束复杂度检测器将在性能规范下最小化两个长度之和。 (iii)最终目标是在VLSI技术中实现检测器。 为了实现这一目标，一个软件检测器，然后硬件检测器将被开发。 这两种探测器将在IBM Almaden研究中心的真实的信道环境中进行测试。