The CFAR method of setting alarms operates by splitting the threshold setting process into two stages.
- Primary Processing adaptively sets a Primary Threshold in the data so that threshold exceedances are maintained at a constant level. This level controls the sensitivity to the defects the symptoms of which are expressed in the data.
- Secondary Processing eliminates uncorrelated exceedances to produce the alert at a known false alert rate; because the alerts from the primary processing are at a known rate then the binary integration process employed in this stage is used to achieve the specified false alert rate.
The strength of the CFAR technique is that it:
- Makes no assumptions about the statistical characteristics of the data.
- Produces alerts that are unaffected by changes in the characteristics of data over time.
- Is impervious to the low frequency changes that are common in HUM data, reacting only to short term effects that relate to a defect.
- It achieves very low false alert rates, to a level where they no longer mask true alerts.
- Can be managed to maintain the target false alert rate without compromising the sensitivity to defects.
The technology provides consistent, robust and reliable alert generation.
