-
Notifications
You must be signed in to change notification settings - Fork 102
/
sp_johansson_impl.m
70 lines (58 loc) · 2.01 KB
/
sp_johansson_impl.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
function [output, fs] = sp_johansson_impl(inp_fs, input, time_skew)
% Reconstruction of a signal from non-uniform samples
% Based on Sindhi and Prabhu's implementation of Johansson's
% method
% index_mapping is used to reorded filters in case the first input
% signal is actually delayed comparing to the second and not vice versa
N = length(time_skew); % number of samplers
fs = inp_fs * N; % full sampling frequency
T_inp = 1/inp_fs;
% index_mapping is used to reorded filters in case the first input
% signal is actually delayed comparing to the second and not vice versa
[time_skew, index_mapping] = sort(time_skew);
x11 = cell2mat(input')';
x11 = x11(index_mapping, :);
TQ = 1; % Nyquist sampling period
% Decimation Periods - in our case all ADCs sample with the same rate
T = [2*TQ 2*TQ];
K = 0.5*lcm(2*T(1), 2*T(2))/TQ; % number of samples in recurrent period
capT = K*TQ; % the full sampling period - of all samplers
M = capT./T;
ML = min(cellfun('length', input)); % number of slices
w_c = 0.85;
LF = lcm(M(1), M(2))*2*K+1; %% min length of LF should be capM*2*K1
n = -(LF-1)/2:1:(LF-1)/2;
taus = time_skew / T_inp * capT; % ADC delays in seconds
tausI = sort([taus(1) taus(2)]);
tauI = zeros(K,ML);
for p = 1:K
tauI(p,:) = tausI(p)+(0:ML-1)*capT;
end;
r = tausI-TQ*(0:K-1);
r = r.';
w_o = w_c*pi*TQ;
hJ = zeros(K,LF);
Nt = (LF-1)/2;
for i = 1:K
C = -2*sin(w_o*(n-r(1+(mod(i-1-n,K)))'))./(pi*(n-r(1+(mod(i-1-n,K)))'));
C(isnan(C))=-2*w_o/pi;
C = C.';
S = zeros(LF,LF);
for k = 1:LF
S(k,:) = sin(w_o*(-Nt+k-1-r(1+(mod(i-1-(-Nt+k-1),K)))-(n-r(1+(mod(i-1-n,K)))')))./(pi*(-Nt+k-1-r(1+(mod(i-1-(-Nt+k-1),K)))-(n-r(1+(mod(i-1-n,K)))')));
end;
S(isnan(S))=w_o/pi;
hJ(i,:) = -0.5*S\C;
end;
x1 = reshape(x11,1,size(x11,2)*K);
y1 = zeros(K,length(x1));
for j=1:K
y1(j,:) = filter(hJ(j,:),1,x1);
y1(j,:) = upsample(downsample(y1(j,:),K,j-1),K)/K;
y1(j,:) = filter([zeros(1,j-1),1],1,y1(j,:));
end;
y = K*0.25*sum(y1,1);
delayJ = (size(hJ,2)-1)/2;
y = real(y(1+delayJ:end));
output = y;
end