Page 497 - Fundamentals of Radar Signal Processing

The threshold T is seen to be proportional to the interference power, ,

with the multiplier α a function of the desired false alarm probability.
To tune the square law detector for a particular radar system, an
acceptable value of P must be chosen. The threshold is then computed
FA
according to Eq. (6.121). The probability of detection that will be achieved is
determined by the target SNR.
Accurate setting of the threshold requires accurate knowledge of the

interference power . In some systems this is known, but in many it is not.

When the interference is principally receiver noise it is possible to measure
and calibrate the detector. In day-to-day operation, however, the receiver noise
will vary over time due to factors such as temperature changes and component
aging. Temperature compensation and periodic recalibration, if possible, can
combat this problem. If the total interference power is significantly affected by
external sources, the variability can be much more severe. In very low noise
radar systems, a significant part of the noise power is cosmic noise. The total

receiver interference then varies with the look direction and the time of day. In
conventional radars, the total interference power can be affected by in-band
electromagnetic interference (EMI). For example, UHF radars can be affected
by television transmissions, while certain wireless communication services can
compete with higher frequency radars, especially in urban areas. If the dominant

interference is ground clutter, its power will vary radically with the type of
terrain being illuminated and even the weather and seasons. For instance, open
desert has a relatively low reflectivity, while refrozen snow can have a very
high reflectivity. Finally, the dominant interference can be hostile
electromagnetic emissions deliberately directed at the radar system (jamming).
In this case, the interference power can be extremely high.
In any of these cases the observed P will vary from the intended value.
FA
To see how significant this variation might be, let P be the intended
FA0
probability of false alarm when the actual interference power is the expected

value of ; thus ln P . Now suppose the actual interference power
FA0
is . The actual P , using Eq. (6.121) with the threshold designed assuming
FA
an interference power of , will be

(6.122)

and the increase in false alarm probability will be a factor of