Page 540 - Fundamentals of Radar Signal Processing
P. 540

c.   Regardless of the LR found in part (b), suppose the rule               is

                           chosen as the detection test. Sketch the resulting curves for P  vs. T
                                                                                                   FA
                           and P  vs. T as T varies from –1.5 to +1.5. Label all important
                                  D
                           values.


                     Problems 3 to 6 are a related group exploring how changes in the values
                     of m,  , and T affect detection and false alarm performance in the real
                     constant-in-WGN example.

                 3.  Consider detection of a real-valued constant in zero-mean real-valued
                     Gaussian noise. Let the noise variance              , the number of samples N = 1,

                     and the constant m = 4. What is the SNR χ for this case? Sketch the
                     distributions p(y | H ) and p(y | H ); label appropriate numerical values on
                                                            1
                                             0
                     the axes. Write the likelihood ratio and log-likelihood ratio for this
                     problem. Simplify the expressions.

                 4.  Continuing with the same parameters given in the previous problem,
                     suppose P  = 0.01 (1 percent) is required. What is the required value of
                                 FA
                     the threshold T? What is the resulting value of P ? Lookup tables or
                                                                              D
                     MATLAB  can be used to calculate the values of functions such as erf(·),
                                 ®
                                  –1
                     erfc(·), erf (·), or erfc (·) that may be needed.
                                                –1
                 5.  Suppose m in Prob. 3 is increased so as to double the SNR;                    and N = 1
                     still. What is the new value of m? Sketch and label the distributions p(y |
                     H ) and p(y | H ) with this new value of m. If the same threshold T found in
                       0
                                       1
                     Prob. 4 is retained, does P  change and, if so, what is the new value? If
                                                    FA
                     that same threshold T is retained, does P  change and, if so, what is the
                                                                     D
                     new value?

                 6.  Go back to the case of m = 4, but now reduce the noise variance to                   . N
                     = 1 still. What is the SNR χ now? Sketch and label the distributions p(y|H )
                                                                                                            0
                     and p(y|H ) with this value of m and  . If the threshold value used in
                                1
                     Probs. 4 and 5 is still retained, what is the value of P ? What is the value
                                                                                    FA
                     of P ?
                          D
                 7.  Consider detection of a constant in complex Gaussian noise. Let the total
                     noise variance          , the number of samples N = 1, and the constant m = 4.
                     What is the SNR c for this case? Suppose P  = 0.01 (1 percent) is
                                                                         FA
                     required. What is the required value of the threshold T? What is the
                                                                                ®
                     resulting value of P ? Lookup tables or MATLAB  can be used to
                                            D
                                                                                                         –1
                                                                                          –1
                     calculate the values of functions such as erf(·), erfc(·), erf  (·), or erfc (·)
                     that may be needed.
                 8.  Compute the threshold T and probability of detection P  for the case of a
                                                                                     D
                     constant in zero-mean complex Gaussian noise, but now with unknown
   535   536   537   538   539   540   541   542   543   544   545