| 摘要 |
A method and apparatus for spectral analysis of a sampled signal are provided. According to the method, a first time interval DELTAT is determined which represents a sample set over which the signal can be averaged for purpose of decomposing the signal into instantaneous mean and random components. From random component of the signal, the optimal time-dependent time interval AW over which the random component should be averaged to obtain the instantaneous frequency spectrum (IFS) is then calculated based upon the instantaneous signal memory. Signal memory specifies the extent in time beyond which the signal "forgets" its past behavior and thus is a measure of its degree-of-randomness. Defining AW in this way reduces the likelihood that short-lived events are lost (or masked) due to uncharacteristically short (or long) averaging intervals. Importantly, both time intervals DELTAT and AW are dynamic and vary randomly with time. Because these intervals can be computed in real time, the instantaneous frequency spectrum over interval AW can also be computed in real time.
|
