Page 519 - Fundamentals of Radar Signal Processing
P. 519

easily  allows  detection  of  both  targets;  the  half  of  the  CFAR  window

               contaminated  by  the  other  target  is  simply  ignored  by  the  SOCA  logic.
               Similarly,  the  SOCA  CFAR  detects  the  target  near  the  clutter  edge,  again
               because the half of the window containing the higher-power clutter is ignored.
                     Figure  6.27a  also  shows  the  principal  failing  of  the  SOCA  method.
               Although  the  CA  CFAR  did  not  exhibit  a  false  alarm  at  the  clutter  edge,  the
               SOCA CFAR does. This is a natural consequence of the SOCA logic. As the

               CFAR window crosses a clutter edge, there will be a region in which the test
               cell  is  in  the  higher  interference  power  region,  while  one  of  the  lead  or  lag
               windows  is  filled  mostly  or  entirely  with  samples  of  the  lower  power
               interference. The SOCA logic ensures that the threshold is then based on the
               lower interference power, significantly raising the probability that the clutter in
               the test cell will cross the threshold.
                     For systems and environments in which closely spaced targets are unlikely

               but the clutter is highly nonhomogeneous, clutter-edge false alarms may be of
               much  more  concern  than  target  masking.  In  this  case,  the  observations  above
               suggest that a greater-of cell-averaging CFAR  (GOCA CFAR) logic be used.
               As  with  the  SOCA  technique,  the  lead  and  lag  windows  are  averaged
               separately, but now the threshold is based on the larger of the two averages:





                                                                                                     (6.154)

               Similar to the SOCA case, the GOCA threshold multiplier is the solution of the
               equation (Weiss, 1982)














                                                                                                     (6.155)

               For N = 20 and              , α  = 7.24.
                                              GO
                     Figure 6.27b illustrates the performance of the GOCA CFAR logic on the
               same  example  used  in Fig. 6.27a.  The  GOCA  threshold  is  now  equal  to  or
               higher  than  the  CA  CFAR  threshold.  Not  surprisingly,  the  GOCA  logic

               successfully  avoids  the  false  alarm  at  the  clutter  edge.  However,  the  strong
               target  masks  the  weaker  target.  Furthermore,  the  weaker  target  even  makes
               detection of the stronger target more marginal, although in this case the detection
               is successful. The GOCA CFAR also misses the target near the clutter edge due
               to the masking effect of the elevated clutter. Additional analysis of the GOCA
   514   515   516   517   518   519   520   521   522   523   524