Welcome Guest, you are in: <root>Elements • Login

Page History: PSD Analyzer Element

Compare Page Revisions

11 10 9 8 7 6 5 4 3 2 1 0 Current 11 10 9 8 7 6 5 4 3 2 1 0

« Older Revision - Back to Page History - Newer Revision »

Page Revision: 2014/12/28 22:39

The element allows to perform a simple power spectrum (PSD) analysis on a real-time signal that is provided by the signal acquisition elements in EventIDE.

Table of Contents [Hide/Show] Actions aligned with epochs of the parent event Properties     Generic Properties Description Snapshots Practical Use     Preparing the element for analysis     Creating the signal plot     Reading the FFT results in code Notes

Description ¶

The signal acquisition elements, e.g. OpenViBE Signal, communicate with recording hardware and buffer incoming data samples at the background of the experiment. If you want to obtain the instant frequency characteristics of a signal, you can use PSD Analyzer. The element operates on a single signal by estimating its power spectrum with the FFT. The estimation is run cyclically in the control loop of the parent event, within a analysis window that embraces the selected number of recent samples. The results include the descriptive statistics over a subset of the signal window: dominant frequency and its power. Other statistics can be done manually, hence the element delivers the FFT output as an array of double numbers. When power spectrum monitoring is required, the element generates a live power/frequency plot, which can be placed on the Status Screen (see the snapshot below).

Snapshots ¶

Status Screen with the PSD plot, frequency statistics and selected powers

Practical Use ¶

Preparing the element for analysis ¶

1. Make sure that your experiment contains one of the data acquisition elements, e.g. OpenViBE Signal element
2. Prepare the data acquisition element, such that it is set to collect at least one signal.
3. Add new PSD Analyzer element to the event, where you plan to process the signal (the element operates only within the parent event).
4. Select the added element to browse its properties in the Property panel.
5. Define a signal for processing, via the Selected Signal property.
6. Adjust the Sampling Rate property such that the value matches to the real sampling rate of the selected signal.
7. Choose the size of the analysis window via the corresponding property.
8. Select the Range Of Interest for statistics and plotting.
9. Define the bin size for the frequency dimension in statistics and plotting.
10. Choose the Power Normalization procedure (the default is normalization for the total signal maximum).
    

Creating the signal plot ¶

1. Create a proxy variable linked to the XAML PSD Plot property
2. Check the plot's visualization settings in the Plot Settings group
3. Add a XAML content control with data binding to the Status Screen code:
   

    
   
   <ContentControl Content='{Binding XAMLPSDPlot}'/>
   
   

   
   

Reading the FFT results in code ¶

1. Create a proxy variable linked to the Frequency Spectrum property
2. Use the Control Loop snippet of the parent event to monitor all signal updates
3. Create a loop to read all signal samples:
   

   
   
   // estimating the spectrum sum
   
   double Sum=0;
   
   for (int i=0;i,FrequencySpectrum.Length;i++)
   
   {
   
   Sum=Sum+FrequencySpectrum[i];
   
   }  
   
   

   
   

Notes ¶

• The Frequency Spectrum values are raw, so you may need to make a conversion to Hz based on the size of analysis window and sampling rate. In contrast, element's statistics and plot automatically scale the FFT results to Hz.
• The dominant frequency and power are searched only inside of the chosen range of interest.
• The frequency spectrum is returned for the entire analysis window, irrelative to the chosen range of interest.
• You can measure a time of one cycle of signal processing with the Control Loop Interval property on the parent event. With plotting on and off, the expected time for a 1000 sample window is ~10 and ~5 ms, correspondingly, given that there are no other processes in the parent event and in the experiment.
• If you are in time-critical section of your experiment, you can temporally cease the signal processing by setting the Is Running' property to false.