Align two arrays by time column

Does anyone know of a MatLab utility to facilitate alignment of two datasets acquired on two devices whose clocks are slightly off? .. we currently resample each and cross-correlate one column from each that contains similar signal (for example the EKG signal) and then shift one of teh two sets .... this is a pain do do manually and i am hoping someone has written a GUI utility to facilitate this ... before we write our own!

[Oleg Komarov]

can you post a brief example of two series and the desired output?

[Philippe]

Table1 (2s sampling rate with missing points):

0 0 14 30,4

2 1,4 14 30,6

4 2,8 14 30,7

6 4,2 14 30,8

8 4,48 14 31

10 7 14 31

12 10 14 31

14 9,8 14 31,1

16 10,4 13 31,3

18 7,56 14 31,4

20 14 14 31,5

22 15,4 14 31,4

24 15,6 13 31,4

26 14,56 14 31,5

28 18,2 13 31,5

30 15,6 13 31,5

32 16,64 13 31,5

34 22,1 13 31,5

36 23,4 13 31,5

38 24,7 13 31,5

40 20,8 13 31,5

42 16,38 13 31,5

44 22,88 13 31,5

46 22,08 12 31,5

48 28,8 12 31,5

50 30 12 31,5

52 24,96 12 31,5

54 32,4 12 31,5

56 26,88 12 31,5

58 19,14 11 31,5

62 20,46 11 31,5

64 28,16 11 31,5

69 30,3 11 31,5

70 21 10 31,5

72 19,44 9 31,5

74 19,98 9 31,5

76 20,52 9 31,5

78 21,06 9 31,5

80 19,2 8 31,5

82 22,96 7 33,7

84 11,76 7 33,9

86 18,06 7 34

88 18,48 7 34,1

90 16,2 6 34,2

92 11,04 6 34,2

94 16,92 6 34,4

96 14,4 5 34,4

98 7,84 4 34,5

100 20 4 34,6

Table2 (3s sampling rate with missing points:

0 7,55390615 9,3871132425

3 26,9564624 35,8988782425000

6 25,7416124 36,5810632425

9 25,9596624 36,7492732425

15 25,8194874 36,6137707425

18 25,65205615 36,5483557425

21 25,66763115 36,4642507425

24 25,7883374 36,3614557425

27 25,58196865 36,4175257425

30 25,7104624 36,4362157425

33 25,5469249 36,3988357425

36 25,6870999 36,5063032425

39 25,5936499 36,3708007425

42 25,7338249 36,3567832425

45 25,50409365 36,4455607425

48 25,63648115 36,4128532425

51 25,64426865 36,4128532425

54 25,3755999 36,4408882425

57 25,4534749 36,3988357425

60 25,63648115 36,3708007425

63 25,63648115 36,3708007425

66 25,4768374 36,2633332425

69 25,57418115 36,4315432425

72 25,5780749 36,3614557425

75 25,5157749 36,3053857425

78 25,5858624000000 36,5249932425000

81 25,5235624 36,4081807425

84 25,58196865 36,2866957425

87 25,49630615 36,1231582425

90 25,4690499 36,2399707425

93 25,58196865 36,2072632425

96 25,5391374 36,2820232425

99 25,5702874 36,3380932425

102 25,4768374 36,2913682425

105 25,55081865 36,2680057425

MergedTable:

0 0 14 30,4 7,55390615 9,3871132425

6 4,2 14 30,8 25,7416124 36,5810632425

18 7,56 14 31,4 25,65205615 36,5483557425

24 15,6 13 31,4 25,7883374 36,3614557425

30 15,6 13 31,5 25,7104624 36,4362157425

36 23,4 13 31,5 25,6870999 36,5063032425

42 16,38 13 31,5 25,7338249 36,3567832425

48 28,8 12 31,5 25,63648115 36,4128532425

54 32,4 12 31,5 25,3755999 36,4408882425

72 19,44 9 31,5 25,5780749 36,3614557425

78 21,06 9 31,5 25,5858624 36,5249932425

84 11,76 7 33,9 25,58196865 36,2866957425

90 16,2 6 34,2 25,4690499 36,2399707425

96 14,4 5 34,4 25,5391374 36,2820232425

Obviously, in pratice there might be no difference in sampling rate or more complicated difference in sampling rate.

Answer 1

[Andreas Goser]

The generic idea is to create a time vector that is equidistant and then map the measured data to it (interpolating). Example code:

a=[1 1; 2 2; 3 2.5; 4 2.2; 5 2.1]; % time and data 'a'
b=[1.1 1.1; 2.1 2.2; 3.0 2.8; 4.1 2.3; 5.0 2.0]; % time and data 'b'
plot(a(:,1),a(:,2),'-ks',b(:,1),b(:,2),'-bs');
t=0:0.2:5; % new equidistant time vector
a_i=interp1(a(:,1),a(:,2),t); % mapping (interpolate data 'a' to new time vector)
b_i=interp1(b(:,1),b(:,2),t); % mapping (interpolate data 'b' to new time vector)
figure
plot(t,a_i,'-ks',t,b_i,'-bs');