Causes
In actual work, it is often necessary to limit the observation time of the signal within a certain time interval. The process of taking a limited amount of data, that is, truncating the signal, is equivalent to multiplying the signal by the window function. If the window function is a rectangular window, the data item will be truncated in the time domain, which is equivalent to the process of periodic convolution of the spectrum of the waveform under study with the spectrum of the rectangular window function in the frequency domain. The distortion spectrum caused by this convolution is shown in Figure 1X2(e^jw). It can be seen that the spectral components are spread out from their normal frequency spectrum, which is called "leakage".
Improvement methods
1. Widen the truncation window and increase the data length;
2. Use a slowly truncated window function to reduce the side lobes of the window function, In order to reduce the leakage energy caused by convolution, such as Hanning window, Hamming window, Kaiser window, etc.
The relationship between spectrum leakage and aliasing distortion
It should be pointed out that spectrum leakage cannot be completely separated from aliasing, because leakage will lead to the expansion of the spectrum, thus making the highest frequency of the spectrum Exceed the folding frequency, causing aliasing.