THE COHERENT
DATA PROCESSING IN RADAR SIGNAL SPECTRAL ANALYSIS WITH SUPERRESOLUTION
Daniel
S. Grigoryan
Radar Data Processing Laboratory
of Army Air Defense Academy, Smolensk
Received
February
23, 2012
Abstract. It is established that at
enough big power of signals, final values of relations a signal/noise in
elements of correlation matrixes don't depend on a way of accumulation of a signal.
Influence of size of a relative interval of the analysis on relative norm of a
indignations vector of weight factors is investigated and its boundary value is
defined. It is defined that at the same order of model relative norm of a indignations
vector of weight factors at use of preliminary coherent data processing in
sample essentially more low, than norm of a vector of indignations of weight
factors at use initial autoregressive methods of the digital spectral analysis.
It is theoretically defined and proved experimentally that efficiency of the
superresolution at identical relations a signal/noise depends on a relative
interval of the analysis without dependence from the chosen method of data
processing. Obvious advantage from the point of view of computing expenses of
coherent methods of a linear prediction with прореживанием under the
superpermission of harmonious signals in comparison with initial methods of the
spectral analysis is shown.
Keywords: superresolution, linear
prediction, weight vector, conditionality, spectral function.