Sequential FFT

Sequential FFT is a technique where the frequency spectrum of the underlying signal is computed by first dividing the signal into overlapping chunks and then computing the FFT-based spectrum of each chunk. The spectrum of the signal is then obtained by averaging the spectra of all the chunks.

This technique is also referred to as Overlap and Add Short Time Fourier Transform (Overlap and Add STFT) or simply Overlapped FFT.

The figure below shows an example of Sequential FFT processing with 50% overlap between sequential chunks of time-domain samples.

Sequential FFT in RFmxSpecAn involves the following steps:

1. The measurement acquires I/Q samples for a duration specified by the sweep time property. The acquired I/Q samples are divided into smaller overlapping chunks of equal size. The size of each chunk is referred to as Sequential FFT Size. In general, Sequential FFT size and RBW filter bandwidth are coupled. Specifying one automatically fixes the other. You can configure the amount of overlap between sequential chunks as a percentage of Sequential FFT Size.
   • When using RFmxSpecAn Spectrum measurement, you can
     • specify the desired FFT size using the Spectrum Sequential FFT Size property with Spectrum RBW Auto property set to True, or
     • specify desired resolution bandwidth by using Spectrum RBW (Hz) property and setting Spectrum RBW Auto property to False; RFmx derives the Sequential FFT Size from RBW filter bandwidth.
   • When using RFmxSpecAn ACP measurement,
     • you must specify desired FFT Size by using the ACP Sequential FFT Size property with ACP RBW Auto property set to True.
     • Setting RBW filter bandwidth is not allowed.
2. FFT spectrum of each chunk is computed by first multiplying the time-domain signal by a window function followed by FFT. Refer to the Window and FFT section for more details.
3. Spectra of all chunks are averaged to obtain spectrum of the underlying signal.

Use Sequential FFT for the following purposes:

1. Fast spectral measurements by utilizing smaller FFT sizes and setting overlap between time-domain data chunks to 0. This ensures that time-domain data chunks are not overlapped, resulting in fewer overall number of FFT operations.
   Note

   • Using smaller FFT sizes can lead to poorer resolution in the spectrum.
   • Eliminating time-overlap between sequential time-domain data chunks can suppress spectral information contained in the beginning and end of each chunk because these parts susceptible to get attenuated by the windowing operation. Sufficient time-overlapping of chunks ensures that no part of the overall signal is completely attenuated by the window operation.
2. Measuring signals with time-varying spectral characteristics. In this case, use sufficient overlap between time-domain chunks. To enable overlapping sequential chunks, set the FFT Overlap Mode property to Automatic (uses 50% overlap) or directly configure the desired overlap using the FFT Overlap (%) property with the FFT Overlap Mode property set to User-Defined.

RFmxSpecAn supports sequential FFT in Spectrum and ACP measurements. You can enable sequential FFT by setting the Spectrum Measurement Method property or ACP Measurement Method property to Sequential FFT. The following properties have limited support in Sequential FFT method.