ABSTRACT
Part III Adaptive Receivers discusses implementation of the third of three
sequentially complementary approaches for increasing the probability of
detection, within at least some cells of the surveillance volume, in presence of
external “noise” which can be either Gaussian or nonGaussian in the spatial
domain but Gaussian in the temporal domain. For each homogeneous subdivision
of the surveillance volume, this approach, identified in the Preface as Approach C,
generally utilizes a nonlinear, nonGaussian receiver whose detection algorithm
is matched to the sampled “noise” voltage spatial probability density function
(PDF) of that subdivision. When the nonGaussian “noise” waveform is spikier
than Gaussian noise, the nonlinear receiver is more effective than a linear receiver
in reducing the detection threshold for a given false alarm rate, provided that the
estimated spatial PDF is an accurate representation of the actual PDF. If the
estimated spatial PDF is Gaussian, then the nonlinear receiver reduces to a linear
Gaussian receiver. The issues are (1) how to model, simulate, and identify the
random processes associated with the correlated “noise” data samples and (2)
how to determine the nonlinear receiver and its threshold which are best matched
to those data samples. Part III Adaptive Receivers addresses these issues and
gives several applications.