Blind System Identification and its Applications

盲系统辨识及其应用

• 批准号: 9710297
• 负责人: Er-Wei Bai
• 金额: $ 28.02万
• 依托单位: University of Iowa
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1997
• 资助国家: 美国
• 起止时间: 1997-10-01 至 2001-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9710297&HistoricalAwards=false
• 关键词: Blind; System; Identification; its; Applications

ECS-9710297 Bai The blind system identification is a variant of the ordinary system identification, in which the input signal to the system is unknown. Despite its wide use in a host of communication and signal processing applications, there is a lack of understanding and efficient schemes . Existing methods of blind system identification can be devided into two categories. The first one assumes some statistical information, usually a whiteness condition, on the unknown input signal, and infeasible approach in many applications since inputs are unknown. The second method is to use training signals and thus reduce the blind system identification to the ordinary system identification. Although simple, this method significantly reduces the efficiency because training signals have sent again and again. For example, in the GSM standard for mobile communication, more than one quarter of the communication capability is used to send training signals instead of sending message signals. This project intends to conduct a thorough study of the blind system identification problem based on our preliminary study in which a new scheme was developed. To the best knowledge of the author, the scheme we propose is the first and in fact the only blind system identification scheme without requiring any a priori information on the unknown input. The proposed research will build on our preliminary results and focus on some fundamental issues as well as on developing algorithms. The expected strategies for conducting such a study are outlined in the proposal.

ECS-9710297 Bai 盲系统辨识是普通系统辨识的一种变体，其中系统的输入信号是未知的。 尽管它在许多通信和信号处理应用中被广泛使用，但缺乏理解和有效的方案。 现有的盲系统辨识方法可以分为两大类。 第一个假设一些统计信息，通常是白度条件，在未知的输入信号，和不可行的方法在许多应用中，因为输入是未知的。 第二种方法是使用训练信号，从而将盲系统辨识简化为普通系统辨识。 这种方法虽然简单，但由于训练信号反复发送，大大降低了效率。例如，在用于移动的通信的GSM标准中，多于四分之一的通信能力用于发送训练信号而不是发送消息信号。 本计画将在前期研究的基础上，针对盲系统辨识问题进行深入的研究，并提出一种新的辨识方案。 据作者所知，我们提出的方案是第一个，事实上是唯一的盲系统识别方案，而不需要任何先验信息的未知输入。 拟议的研究将建立在我们的初步结果，并专注于一些基本问题，以及开发算法。 该提案概述了开展这一研究的预期战略。