Find the start and end position of a data frame in a pulsed signal

Hello everyone,

I have a signal like this:

and would like a good way to find the start and end of each frame. Note that the size of each frame can be variable and sometimes the flat part in between could have an offset.

Thanks for your time and effort,

Paul

Answer 1

[Joseph Cheng]

so what one would do first is threshold a signal and perform an edge detection on it. something like this:

close all;
figure(1)
x = 0:.001:10;
y = 10*sin(50*pi*x);

bduration = [996   830   449   898   680 200];
bpoints = [1        3275        4582        5758        8211 9800];
for ind = 1:length(bduration)
    y(bpoints(ind):bpoints(ind)+bduration(ind)) = 0;
end
y = y+(2*rand(1,length(x))-1);

minL = 160; %minimum samples for signal burst.
Ty = zeros(size(y));
Ty(abs(y)>5) = y(abs(y)>5);
edgefilt = [ones(1,minL) -ones(1,minL )];

Edges = [zeros(1,minL-1) conv(abs(Ty),edgefilt/160,'valid') zeros(1,minL)];

plot(x,y,x,Edges,'r');

and as you can see from the final plot just find the locations of the peaks.

[Paul Vicioso]

Hi Joseph, thanks for your answer, and now it is partially working, but I have a problem detecting the edges of small signals, for example, in the next graph:

I can detect the edges of the first few frames but not (consistently) the smaller ones (to the right), can you give me a hint of how to do it or even how to tweak your function to detect these signals?

Thanks for your help,

Paul Vicioso