EXFO FTB-5600 Distributed PMD Analyzer - Manual (Page 136)

Optimizing Measurements
130 FTB-5600
Use an Appropriate Spatial Smoothing Filter Value
³
A median filter provides the medium of a group of points. The median
filter will preserve sharp transition, whereas the moving average filters
will not.
In the figure below, you can see that the sharp transition is still
preserved and that there are no delays.
When filtering a cumulative PMD curve, you must take into account the
behavior of the filters. You might have to adjust the positions of the markers
after you have applied a filter.
In the figure below, you can see how the filter behaves when encountering
a small and a large step compared to the noise variance. The red curve is
the anti-causal median filter and the blue curve is the causal moving
average.
In the case of the small step, the moving average and the median filtering
show the same behavior, that is a ramp equal to the filter length to reach
the top of the step.