create_PHI

 Inputs
   P -- desired number of sampling windowsLOT window
   N -- length of each sampling window
   M -- number of measuremnets per sampling window
   mType -- type of measurements (randn, sign)
   SM -- sampling mode (starting and ending position of each row)
       1 : Project sampling waveforms on the LOT basis 
           Sample each window separately
           Resultant system will be block diagonal with LOT coeffs
           measureed with mType waveforms
       2 : Project sampling waveform on the LOT basis
           Sample multiple windows in each measurement
           Resultant system is banded 
       3 : General sampling system
           Sample signal directly with mType samples
           Resultant system will be banded with LOT coeffs. measured with
           mType*PSI waveforms... 
       4 : Overlapping samples
           Applicable with compressive interference cancellation
       5 : A windowed measurements and reconstructed with DCT-I basis at 
           right-side border

0001 function PHI = create_PHI(in);
0002 
0003 % Inputs
0004 %   P -- desired number of sampling windowsLOT window
0005 %   N -- length of each sampling window
0006 %   M -- number of measuremnets per sampling window
0007 %   mType -- type of measurements (randn, sign)
0008 %   SM -- sampling mode (starting and ending position of each row)
0009 %       1 : Project sampling waveforms on the LOT basis
0010 %           Sample each window separately
0011 %           Resultant system will be block diagonal with LOT coeffs
0012 %           measureed with mType waveforms
0013 %       2 : Project sampling waveform on the LOT basis
0014 %           Sample multiple windows in each measurement
0015 %           Resultant system is banded
0016 %       3 : General sampling system
0017 %           Sample signal directly with mType samples
0018 %           Resultant system will be banded with LOT coeffs. measured with
0019 %           mType*PSI waveforms...
0020 %       4 : Overlapping samples
0021 %           Applicable with compressive interference cancellation
0022 %       5 : A windowed measurements and reconstructed with DCT-I basis at
0023 %           right-side border
0024 
0025 % For SM = 1,
0026 % In the LOT coefficients domain, the final system matrix will be block
0027 % diganal, where each block measures only one set of LOT coefficints.
0028 % In the signal domain, the measurement waveforms are created by first
0029 % projecting random (Gaussian or Bernoulli) vectors onto LOT bases...
0030 % The sensing matrix rows will overlap and align with the boundary of each
0031 % LOT window boundaries
0032 % LM = N+2*eta*N; length of sensing waveform
0033 % Note: Each set of LOT coefficients can be estimated separately.
0034 
0035 % For SM = 2, (multiple LOT coefficients measured together--streaming setup)
0036 % In the LOT coeffs domain, overlapping measurements of LOT coefficients
0037 % in d windows
0038 % LM = d*N+2*eta*N; length of sensing waveform in the signal domain
0039 % Note: Need streaming setup here
0040         
0041 % For SM = 3;
0042 % 'universal' sampling scheme
0043 % It might be better to align the measurements such that they overlap with
0044 % a few LOT windows before the last overlapping interval
0045 % LM = d*N; length of sensing waveforms overlapping d LOT windows
0046 
0047 % Sensing matrix for each
0048 genAmat_h = @(M,N) genAmat(M,N,in);
0049 
0050 P = in.P; M = in.M; N = in.N; LM = in.LM;
0051 PHI = zeros(P*M,(P-1)*N+LM);
0052 for p = 1:P
0053     PHI((p-1)*M+1:p*M,(p-1)*N+1:(p-1)*N+LM) = genAmat_h(M,LM);
0054 end