The output of the monitor is a
set of time series representing the running median estimate of the
second moment for different frequency bands. These estimates are
calculated for different channels. Furthermore, the monitor computes
the standard deviation sigma of white noise by a Monte Carlo simulation
with a Gaussian distribution N(0,1). Every time an estimated second
moment crosses the three sigma threshold, a trigger is recorded and the
data is inferred to have departed from the stationary behaviour. The
monitor accepts many parameters which have to be specified in a
configuration file. Some of these parameters are the size of the time
stride, the number and values of the frequency bands and the channels
to be processed. Additionaly the monitor produces a 12 hour data trend
plot.
The result files record four quantities : the minimum, maximum, rms of
the output and percentage of the data points crossing the threshold in
a one hour period. Those segments that record more than 25% threshold
crossing need further attention and are useful for construction of data
quality flags.
|