In general, sure. Noise can be white noise (totally random), pink noise (band limited), motors, variac dimmers, it can be interference from other signals in the same channel, or it can be intersymbol interference (ISI), which is the signal interfering with itself, as it travels over multiple paths from source to destination. It's the sum of all of these.

If you subtract the s(t), and let the receiver receive what's left, it would get a good idea of what racket there might be in that channel. Use a spectrum analyzer, for instance, and see how flat the noise is, or isn't. Although that would not predict ISI for the desired s(t).

@ Albert Manfredi

Thanks for your reply, my main concern is now:

If I would like to find the probability of detection for some conventional energy detector, Is it really matter to take the noise uncertainty in consideration, If I was measuring the the incoming noise directly from the received signal?

Some types of noise can be more disruptive than others, for the receiver. I think this is what you're asking. For example, Gaussian noise is less disruptive, for a give noise power, than would be co-channel interference from another signal source? Or on the other side, ISI would be less disruptive than Gaussian noise, if you have a receiver with good feedback equalizer, capable of twisting the known symbols back in shape.

So I'd say, in general, not all noise is made equal.

Noise uncertainty means inaccurate estimation of noise power. if u have used randn for n(t), Multiply that with noise power.

for example if we assume noise uncertainty of -90 dBmW , that means we are generating noise with that noise power. 

convert it into linear scale 

noise_uncertainity = [-90*10^-3] ; %noise power in db

np=10^(noise_uncertainity/10);

n(t)=np*randn(1,N);

r(t)=s(t)+n(t)

i think you are clear now

@ Swetha Namburu

Thanks very much

dear mahdi if we know the noise power no problem in communication systems and every think is simple, but in actual the noise power or in general noise behavior is random and we must dealing with this random by statistical theory such pdf ,mean expected value and sigma . to solve this problem match filter is optimum solution which take all signal power with minimum noise .for more see match filter transfer function .

Dear Mntaser Saleem

Thanks very much for your feedback.

Best,

In case of a Wireless channel, The model will become r(t) = h(t)*x(t) + n(t) where h(t) is the impulse response of the channel and n(t) is any additive noise (either internal electronics or external interference) . 

you can generate h=randn(1,taps) + j*randn(1, taps)

in addition to what is suggested by Sweta.

Dear Syed Muslim Shah, 

Thanks for your clarification.

Best,