gaussian noise in infinite samples in real time matlab

I am running a real time experiment which is always running and will stop if a user will send stop command.

The gaussian noise has to be added for infinite samples till the time user is not stopping the simulation.

How can I add gaussian noise for infinit samples.

[Mathieu NOE]

hello

look at methods for recursive random number generators

Good Parameters and Implementations for Combined Multiple Recursive Random Number Generators (umontreal.ca)

or use the pseudo random binary sequences

Generate pseudo random numbers problem [ SOLVED ] - MATLAB Answers - MATLAB Central (mathworks.com)

[Rashi Mehrotra]

Thanks. The solution given in the link is not for infinite samples, but yes it gives the solution for updating the counter. Please if you can tell how to design noise for infinite samples and stop on the user command.

[dpb]

Your Q? is far too imprecise to know of what you're asking for, but simply using randn each time need a new sample is almost surely all that is needed to be done.

The default stream in MATLAB has an approximate periood of 2^19937 - 1 which is a close-enough representation of infinite that will surely outlast the user's attention span. randn uses the same rng stream as does rand

Explain what is the real question here.

[Rashi Mehrotra]

Th real question is I have to add gaussian noise for every sample of data which is received by the radar unit. This will add gaussian noise for infinite sample and stop whenever the radar stop sending. I hope it is clear now. Should I take the value of N as 2^19937 - 1 or for the infinite samples some other function has o be made.

Answer 1

[Chunru]

t = 0;

sigma = 1; % noise rms value

while true

x = cos(2*pi*0.01*t); % received data

x = x + sigma * randn; % add noise

t = t + 1;

end

This way, you will keep adding noise to your received data indefinitely. The random number generator in matlab may repeat after some numbers ( 2^19937 - 1 ). This will be a huge number. If you using a sampling freuency of 1MHz, it may take this number of years :

vpa(sym(2)^19937/(1e6*365.25*24*3600))

ans = 

1.367475599344741884064452149827e+5988

This wil take 10^5988 years (can you imagine how big this number is?) to repeat the random number. So you don't have to worry when the random number will repeat.

Anyway, even if it repeats, it unlikely has any side effect to your simulation.