model.xml

<project name='project_01' pubsub='auto' threads='1' use-tagged-token='true'> 
<description> 
	This example project is a basic demonstration of using the Short-Time
	Fourier Transform (STFT) algorithm. A Fourier transform decomposes a
	function of time into its underlying frequencies. The amplitude, offset,
	and rotation speed of every underlying cycle is returned by the function.
	STFT is commonly used to monitor the time-varying frequency content of a
	signal. It can be used in digital filters to detect anomalies in continuous
	streams of data.  

	This project contains a single continuous query consisting of the following: 
	1.	A source window that receives the data to be analyzed 
	2.	A calculate window that calculates STFTs on incoming data and publishes
		the results
</description>
<contqueries>
  <contquery name='cq_01' include-singletons='true' trace='w_calculate'>
   <windows>
    <window-source name='w_source' insert-only='true' autogen-key='true' index='pi_EMPTY'> 
     <description> 
	This source window receives the included example input data file via a file
	socket connector.The input stream is placed into two fields for each observation:
	an ID that acts as the data stream's key, named ID; and a y coordinate of data
	named y.
     </description>
     <schema>
      <fields>
       <field name='ID'  type='int64' key='true'/>
       <field name='y' type='double'/>
      </fields>
     </schema>
     <connectors>
      <connector class='fs' name='publisher'>
       <properties>
        <property name='type'>pub</property>
        <property name='fstype'>csv</property>
        <property name='fsname'>input.csv</property>
        <property name='transactional'>true</property>
        <property name='blocksize'>1</property>
        <property name='addcsvopcode'>true</property> 
        <property name='addcsvflags'>normal</property> 
        <property name='noautogenfield'>true</property>
       </properties>
      </connector>
     </connectors>
    </window-source>
    <window-calculate name="w_calculate" algorithm="STFT" index='pi_EMPTY'>
      <description> 
	The calculate window, w_calculate receives data events from the source window. It
	publishes calculated transforms according to the STFT algorithm properties. In this
	example, the following algorithm parameters govern the STFT algorithm: 

	windowLength:		Specifies the length of the sliding window. The value you 
				specify must be greater than the value of overlap. 
	windowType:		Specifies one of the following window types: 1=Bartlett,
				2=Bohman, 3=Chebyshev, 4=Gaussian, 5=Kaiser, 6=Parzen,
				7=Rectangular, 10=Tukey, 11=Bartlett-Hann, 12=Blackman-Harris,
				13=Blackman, 14=Hamming, 15=Hanning, and 16=Flat Top. 
	fftLength:		Specifies the length to which window data should be expanded.
				Zeros are appended to the data before the Fast Fourier Transform
				(FFT) is performed. The specified value must be positive and at
				least as large as windowLength. To maximize computational efficiency
				it is suggested that a power of two is specified.
	binsInSchema:		Specifies the number of frequency bins to output. This cannot
				exceed the value of fftLength. For real signals, bins greater than
				(fftLength/2) are not physically meaningful. 
	overlap:		Specifies the overlap between consecutive windows. This must be less
				than windowLength.
	
	Additionally, the following input-map and output-map properties define the STFT algorithm in
	this calculate window: 
	
	input:			Specifies the input variable by its name in the source schema. The
				calculate window will analyze this variable.
	timeId:			Specifies the time ID variable name in the input stream. This variable
				should be of type int64. 

	timeIdOut:		Specifies the time ID variable name in the output stream. There will be
				more than one output event for a given time ID.
	binOut:			Specifies the frequency bin variable name in the output stream. 
	powerOut:		Specifies the name of the power variable in the output stream.
	phaseOut:		Specifies the name of the phase variable in the output stream. 

	For this example, the output mapping is using a non-list format. An output-mapping that uses a list
	output is also supported for this algorithm. 

	The resulting calculated transforms are written to the file result.out via a file socket connector
      </description>
      <schema>
        <fields>
          <field name='time' type='int64' key='true'/>
          <field name='bin' type='int64' key='true'/>       
          <field name='power' type='double'/>
          <field name='phase' type='double'/>
        </fields>
      </schema>
        <parameters>
          <properties>
           <property name="windowLength">64</property>
           <property name="windowType">12</property>
           <property name="fftLength">256</property>
           <property name="binsInSchema">256</property>
           <property name="overlap">32</property>
          </properties>
        </parameters>
        <input-map>
          <properties>
           <property name="input">y</property>
           <property name="timeId">ID</property>
          </properties>
        </input-map>
        <output-map>
          <properties>
           <property name="timeIdOut">time</property>
           <property name="binOut">bin</property>
           <property name="powerOut">power</property>
           <property name="phaseOut">phase</property>
          </properties>
        </output-map>
     <connectors>
      <connector class='fs' name='sub'>
       <properties>
        <property name='type'>sub</property>
        <property name='fstype'>csv</property>
        <property name='fsname'>result.out</property>
        <property name='snapshot'>true</property> 
   	<property name='header'>true</property>
       </properties>
      </connector>
     </connectors>
    </window-calculate>
   </windows>
   <edges>
    <edge source='w_source' target='w_calculate' role='data'/>
   </edges>
  </contquery>
</contqueries>
</project>