Abstract: Cohen's bilinear time-frequency distribution has received considerable attention as a powerful tool for analyzing nonstationary signals. In this paper, a new discrete-time time-frequency distribution is introduced. Unlike the existing analytic signal method, the new one utilizes all signal local autocorrelation information, so that, without computing analytic signal, one can avoid aliasing when the signal is sampled at one Nyquist's rate. Some desirable distribution properties and associated kernel requirements are also discussed.

Key words: time-frequency representation, signal analysis, nonstationary signal analysis

摘要： Cohen类双线性时频分布作为一种有力的非平稳信号解析工具，得到日益广泛的应用。该文引入了一种新的双线性离散时间时频分布，与原有的利用解析信号方法相比，它利用了信号全部局部自相关信息，从而在不必计算解析信号的情况下，解决了按一倍Nyquist率对原始信号抽样时其时频分布产生混叠的问题，并且讨论了新方法的性质及其核函数应满足的条件。

关键词: 时频表示, 信号分析, 非平稳信号分析