Page 70 - Autonomous Mobile Robots
P. 70

Millimeter Wave RADAR Power-Range Spectra Interpretation    53

                              (b)
                                 30

                                 25

                                 20

                                 15
                                Power (dB)  10



                                  5

                                  0

                                 –5

                                –10

                                        7      8      9     10     11      12     13     14
                                                            Range (m)

                              FIGURE 2.5 Continued.

                              the power variance of the noise at the peaks is less than that in the rest of the
                              signal.

                              2.4.3 Noise Analysis during Target Absence and
                                    Presence
                              As indicated in Figure 2.5b, the power noise variance is different at the power
                              peaks (target presence) and nonpeaks (target absence) sections of the power
                              range spectra. Therefore, the noise statistics at the RADAR receiver outputs
                              during target absence and presence will now be derived. Knowledge of the
                              noise distributions is necessary for accurately predicting the RADAR spectra
                              for prediction of feature location in robot navigation.

                              2.4.3.1 Power-noise estimation in target absence
                              The noise in the voltage signal entering the mixer stage is assumed to be
                              Gaussian distributed with zero mean. A theoretical analysis to determine the
                              power-noise distribution, after this signal has passed through the low pass and
                              high pass filter stages, and the FFT process has been given in Reference 15.
                              However, due to the unknown nature of the exact internal components within




                              © 2006 by Taylor & Francis Group, LLC



                                 FRANKL: “dk6033_c002” — 2006/3/31 — 17:29 — page 53 — #13
   65   66   67   68   69   70   71   72   73   74   75