Page 159 - Fundamentals of Radar Signal Processing
P. 159

a three-dimensional convolution of the effective reflectivity with a convolution

               kernel comprising the antenna two-way voltage pattern in the angle coordinates
               and the pulse modulation function in the range coordinate. Specifically,




                                                                                                     (2.116)

               where  the  symbols  ∗,  ∗ ,  and  ∗   denote  convolution  over  the  indicated
                                                          ϕ
                                               θ
                                           t
               coordinate.  Now  assume  the  antenna  pattern  is  symmetric  in  the  two  angular
               coordinates, as is often the case; rescale the time variable to units of range; and
               replace θ   and ϕ   with  general  angular  variables θ  and ϕ. These substitutions
                                   0
                          0
               finally give





                                                                                                     (2.117)

                     Equations  (2.116)  and (2.117)  are  stated  as  approximate  convolutions
               because of the finite integration limits in the angular variables, which arise due
               to the periodicity in angle of the antenna pattern and scene reflectivity. A full
               discussion  of  a  spherical  convolution-like  equation  similar  to Eq.  (2.114),

               including development of the Fourier transform relations, is given in Baddour
               (2010). Nonetheless, like a linear convolution, Eq. (2.117) computes the output
               at  a  given  point  in  space  as  a  local  average  of  the  reflectivity  distribution,
               weighted by the antenna pattern and waveform. For most antennas and pulses,
               these patterns concentrate most of their energy in a relatively small finite region

               defined by the mainlobe for the antenna pattern and the pulse duration for the
               waveform. Consequently, the output signal can be expected to behave like a true
               linear convolution.
                     The  convolutional  model  of Eq.  (2.117)  is  an  important  result.  Its
               significance is that it allows interpretation of the measured data as the result of a
               linear filtering process, so that Fourier transform relations between y(θ, ϕ, R),
                               2
               ρ′(R, θ, ϕ), E (θ, ϕ),  and x(t) can be established and applied to model signal
               properties,  determine  sampling  rates,  and  so  forth.  For  example,  the  range
               resolution of the measured reflectivity function is seen to be limited by the pulse
               duration.  (In Chap. 4 it will be seen that the introduction of matched filtering
               will significantly change this statement.) Similarly, for a conventional scanning
               radar, the angular resolution will be determined by the antenna beamwidth. (In

               Chap. 8 it will be seen that the introduction of synthetic aperture techniques also
               significantly changes this statement.) It also follows from the filtering action of
                            2
               x(t)  and E (θ,  ϕ)  that  the  bandwidth  of  the  measured  reflectivity  function  in
               range  and  angle  is  limited  by  the  bandwidth  of  the  waveform  modulation
               function and antenna power pattern. This observation will be used in Chap. 3 to
   154   155   156   157   158   159   160   161   162   163   164