BalahariVignesh/Radar-Target-Generation-and-Detection

Radar-Target-Generation-and-Detection

Range from First FFT

Range Doppler Response

Implementation steps for the 2D CFAR process.

1. Selection of Training, Guard cells and offset.

• The training cells, Guard Cells and Offset are chosen after different trials based on the scenario.
• When offset is too small, it is unable to seperate the signals from noise
• When Training Cells is too small, we keep missing the actual target signal.
• When Training Cells is too large, it is unable to find the target signal and we get a plane.
• When Guard Cells are too smal, ther is very little leakage of target signal.
• When Guard Cells are too high, there is leakage to other bins.

  %Select the number of Training Cells in both the dimensions.
  Tr = 16; % 12 16 10
  Td = 8; % 6 8 4
  %Select the number of Guard Cells in both dimensions around the Cell under
  %test (CUT) for accurate estimation
  Gr = 5; %5 8
  Gd = 2; %2 4
  % offset the threshold by SNR value in dB
  offset = 7;
  %The training cell size without guard cells
  TrainingCells = (2*Tr+2*Gr+1)*(2*Td*Gd*1)-(2*Gr+1)*(2*Gd+1);

2. Steps taken to suppress the non-thresholded cells at the edges.

• Two matrix are declared of same size as RDM, to hold threshold values(threshold_cfar) and the final output signal(Signal_cfar).

  % Vector to hold threshold values
  threshold_cfar = zeros(r,c);

  %Vector to hold final signal after thresholding
  signal_cfar = zeros(r,c);

• The threshold is calculated by traversing along training cells.
• The total threshold is calculated and the values of guard cells are later subtracted.
• Once the noise level is calculated for training cells, the average is calculated by dividing the values by number of training cells.
• Offset is added to the noise, also converting back it to logarithmic scale using pow2db.
• The signals under CUT are compared against threshold.
• If CUT level>threshold assigned a value 1.(I am not assigning 0 as they are already a zeros matrix same of RDM size.)

  %Cell Under Test

  for i = (Tr+Gr+1):(r -2*Tr-2*Gr)
    for j = (Td+Gd+1) : (c - 2*Td-2*Gd)
        %Summing the signal level across the complete grid.
        threshold_cfar(i,j) = sum(sum(db2pow(RDM(i-(Tr+Gr) : i+(Tr+Gr),j-(Td+Gd) : j+(Td+Gd)))));           
        %Now subtracting the Guard Cells noise from the calculated noise sum of whole grid.
        threshold_cfar(i,j) = threshold_cfar(i,j) - sum(sum(db2pow(RDM((i-Gr):(i+Gr),(j-Gd):(j+Gd)))));
        %taking the average of noise
        threshold_cfar(i,j) = threshold_cfar(i,j)/TrainingCells;
        %Adding offset and converting back to log scale.
        threshold_cfar(i,j) = offset + pow2db(threshold_cfar(i,j));
        %Compare CUT(RDM(i,j) against threshold_cfar
        if(RDM(i,j) > threshold_cfar(i,j))
            signal_cfar(i,j) = 1;
        %else
            %signal_cfar(i,j) = 0;
        end
    end

  end

Final CFAR Output