Page 131 - Radar Technology Encyclopedia
P. 131

detection, noncoherent                                                        detection probability  121



               The optimum detector for a noncoherent pulse train typi-  where L  is the detection threshold; f (y,u) and f (y,0) are the
                                                                       0
           cally consists of the matched filter (matched to single pulse in  probability density functions of the received signal  y(t) in
           a train), an amplitude detector, a noncoherent integrator, and a  interference when the desired signal  u(t) is present in the
           threshold unit (Fig. D25). The dependence of the signal-to-  observation interval T and when it is absent, respectively. The
           noise ratio versus the number of pulses required for detection  likelihood ratio determines by what factor the observed pres-
           with specified probabilities of detection and false alarm can  ence of u(t) is more probable in the actual presence of a signal
           be determined from detection curves for chosen target-fluctu-  than during its absence.
           ation model and detector model. AIL, SAL                 The decision “there is a signal,” given by the detector, is
           Ref.: Blake (1980), p. 41; Sosulin (1992), p. 64.    written in the form R[u(t)] = 1 and the decision “there is no
                                                                signal” is given in the form R[u(t)] = 0. As a result we can
                                                         Alarm
                 Matched    Envelope      Video   Threshold     write
                     filter  detector  integrator
                                                                                         1 L ³,  L 0
                                                                                Rut ()[  ] æ     ö
                                                                                       =
              Figure D25 Optimum detector for a noncoherent pulse train.                 è 0 L <,  L ø
                                                                                                0
                                                                    In place of the likelihood ratio L sometimes one uses the
           nonparametric detection (see distribution-free detection,
                                                                logarithmic function of this ratio
              CFAR).
                                                                                             fy u )
                                                                                               ,
                                                                                             (
           Open-loop automatic detection is used in an ESM system in           l =  log  L =  log  ---------------
                                                                                               ,
                                                                                             fy 0 )
                                                                                             (
           which a certain number of many cells are available for incom-
                                                                In this case the decision rule for detection is  written in the
           ing  pulses (Fig.  D26). The available cells are labeled with
                                                                form
           information about the instantaneous parameters of the emis-
           sion (i.e., direction-of-arrival or pulsewidth) and when a pulse             1 l ³,  log L 0
                                                                                      =
           arrives, a free cell will declare itself available to accept other  Rut ()[  ] æ       ö
                                                                                        è 0 l <,  log L ø
                                                                                                0
           pulses with the same parameters as those of the first pulse,
                                                                Such detection  frequently is called  logarithmic. There is a
           within a certain tolerance. The agility of the parameters may
                                                                large number of variants for detection of different signals in
           be taken into account to use the appropriate criterion. SAL
                                                                the background of different kinds of interference, and in each
           Ref.: Neri (1991), p. 309.
                                                                case the likelihood ratio or its algorithm can be specified for
                  PW, DOA, frequency  Detection  Analysis, tracking,
                          histogram          and identification  known characteristics of interference and mixture of a signal
                                                                with interference  to find  the optimum  detection  algorithm.
                     Cell 1                   Association
                                 Verification  of new           Optimum detection is discussed in radar detection theory for
                                   of time     emitters         finding maximum achievable signal detection performance in
                                  relation      with
                                  between  Emitter  old ones;   a  background of interference. In practice  it is  difficult  to
                                 the pulses   frame  comparison
           Front           Pulse    in each   with library      implement completely optimum  detection algorithms, and
            end            frame
                                 histogram    and updating      simplified detectors with parameters close to optimum are
                                     and      of track file
                                 automatic                      used (e.g., the  binary detector,  m-out-of-n detector, etc.).
                                 detection
                                               Library          These detectors experience some detector loss with respect to
                     Cell N                     Track           the true optimum and  are  termed  quasioptimum detectors.
                                                file
                                                                AIL
             Figure D26  Open-loop automatic  detection (from Neri,  1991,  Ref.: Skolnik (1970), p. 15.5; Dulevich (1973), pp. 60–62; Blake (1980),
             Fig. 4.27, p. 310).                                   p. 45.
           Operator detection is the detection of echo signals by the  Detection probability is “the probability that a signal, when
           human operator using visual  observation of radar displays  actually present at the input to the receiver, will be correctly
           (sometimes called  visual detection). Operator detection  declared a target signal based on observation of the receiver
           depends not  only on the  signal-to-noise  ratio but  on  the  output.” This probability is typically referred to in terms of
           visual-mental acuity of the operator, his alertness and experi-  single-pulse, single-scan, or cumulative probability of detec-
           ence, and the resolution of the display (which may cause a  tion. Single-scan probability of detection is often called the
           collapsing loss).  In  modern radars this  method is usually  blip-scan ratio, for which the usual notation is P . For sur-
                                                                                                         d
           replaced with automatic detection. SAL               veillance radars, the probability that the target is detected at
           Ref.: Barton (1964), p. 6; Skolnik (1990), p. 2.24.  least once in N successive scans is called the cumulative prob-
                                                                ability of detection.
           Optimum detection is detection of a signal in an interference
                                                                    The probability of detection is associated with the proba-
           background using optimum detection criteria. In the optimum
                                                                bility of false alarm in a way that depends on the detection
           detection a decision on presence of a target is made in the
                                                                criterion. For a given receiver configuration if the detection
           case when the likelihood ratio
                                                                threshold is reduced to increase the probability of detection,
                                     ,
                                   fy u )
                                    (
                              ,
                                  =
                             (
                            L yu ) --------------- ³ L 0        there will be an increase in false-alarm probability, and simi-
                                    (
                                     ,
                                   fy 0 )
                                                                larly for decreased probabilities if the threshold is raised. The
   126   127   128   129   130   131   132   133   134   135   136