DMT tool for production of burst FOMs
S.Klimenko and A.Sazonov ,
University of Florida
BurstMon is the DMT tool for monitoring of the performance of LIGO
detectors. The performance estimation is based on the burst Figures of
Merit (BFOMs) produced in real time for each LIGO detector. There are
three types of BFOMs: 1) glitch rates, 2) detector sensitivity to
injected waveforms and 3) noise variability. The monitor produces
the DMTViewer and trends: 1 min trends for rate and sensitivity,
and 1 sec trends for noise variability.
- rates: raw rates after
cluster reconstruction at given black pixel probability (1% by default).
- sensitivity: For
estimation of the detector sensitivity, the BurstMon performs a real
time simulation by injecting simulated waveforms of different strength
into the AS_Q data. The waveforms are provided by user. The injected
signals are detected with the
algorithm similar to WaveBurst, but working at one specific time-frequency
resolution. The detection efficiency is reconstructed for each type of injected
waveforms. The detector sensitivity is estimated as the root-square-sum
amplitude at 50% detection efficiency. Currently the BM produces uncalibrated sensitivity.
- variability: The
BurstMon tracks variability of the noise in selected frequency bands.
The noise variability is calculated in three steps: 1) perform wavelet
transformation to obtain TF plot with resolution of 16 Hz x 1/32
sec, 2) whiten
data - normalize each wavelet layer by the noise RMS in this layer
(averaged over BM stride) and 3) calculate the RMS of the normalized
wavelet coefficients with the same time stamp - this is the
noise variability. It is calculated every 1/32 sec and it is
supposed to be close to unity for stationary noise.