Table 4 Procedure for the iterative sidelobe removal
From: A frequency-sharing weather radar network system using pulse compression and sidelobe suppression
3-a. | Start with the MF output \(\boldsymbol y_{MF}^{(k)}\) in (22) as the input. |
Set the processing output y (k) to 0 initially. | |
3-b. | Set the sequence a for the inner-loop processing as \(\boldsymbol a=\boldsymbol y_{MF}^{(k)}\). |
3-c. | Find the peak of |a|: specifically, \(i_{M}=\arg \max \limits _{i} \left | a[\!i] \right |\) and M=a[i M ]. |
Let a 0 denote the sub-sequence of a with length 2L−1, which is centered at i M : specifically, a 0[ i]=a[i M +i] for i=−L+1,−L+2,⋯,L−1. Then, evaluate the sidelobe level \(Q\left (\boldsymbol a_{0}\right)=\sum \limits _{i \neq 0} \left |a_{0}[\!i]\right |\) of a 0. | |
Set the weight β for sidelobe subtraction to the initial value β 0∈(0,1). | |
3-d. | Tentatively subtract a portion of the sidelobe component corresponding the detected peak from a 0, and denote the result by a 1: specifically, a 1=a 0−β M R 1. Then, evaluate the the sidelobe level Q(a 1) of a 1. |
3-e. | If |Q(a 0)−Q(a 1)|<γ β M Q(R 1) for the threshold coefficient γ∈(0,1], go to 3-f1. Otherwise, go to 3-f2. |
3-f1. | Replace β with β/2. If β=β 0/32, stop the trial of sidelobe subtraction for this peak and go to 3-g. Otherwise, go back to 3-d. |
3-f2. | Accept the sidelobe subtraction as follows: a[ i M +i]=a[ i M +i]−β M R 1[ i], \(y_{MF}^{(k)}[\!i_{M}+i]=y_{MF}^{(k)}[\!i_{M}+i]-\beta M R_{1}[\!i]\), and y (k)[ i M +i]=y (k)[ i M +i]+β M R 1[ i]. |
3-g. | Make this peak unavailable as a peak again in any subsequent inner-loop by setting a[i M ]=0. |
3-h. | If the average amplitude \(\frac {1}{L_{T}} \sum \limits _{i} \left |a[\!i]\right |\) of a is smaller than the preset value, quit the inner loop and go to 3-i. Otherwise, go back to 3-c. |
3-i. | If the standard deviation of \(\boldsymbol y_{MF}^{(k)}\) is close to the expected noise level, stop the procedure and output y (k). Otherwise, go back to 3-b. |