power spectral density clarification required

[Y.SAI SARASWATHI]

Hello,
why psd is considered in every communication system ,why is it so significant ,can any one explain me about its practical significance(please avoid math analysis)
thank you.

 

[u7karsh]

For any signal, it is the frequency components that matters.. basically the shape of wave in frequency domain matters.. time domain representations are not considered since even if the signal is attenuated in voltage, it can still be recovered by amplifying it. Thus a more solid representation is needed. In frequency domain, we see the power of each frequency component of interest. If the wave is very much attenuated but still the power of actual wave's frequency component is more than the superimposed noise, we can still recover it (or an aprroximate of it) Thus we consider PSD over time domain representation. Hope this helps.. More to it in time domain you would say that noise must not be there at all while if in frequency domain, if we know that we have a low pass signal and noise is of high frequency, it will never distort my low pass signal.

 

[Y.SAI SARASWATHI]

Thank you,
i need some more clarity,
frequency domain is just for theoretical analysis wright? then if a signal is affected by noise it also means that its frequency components are also affected and consider a wireless channel where we cannot expect kind of noise then how are we able to receive signals correctly. (so confused)

 

[u7karsh]

Both frequency and time domain are inter related.. it is obviously for both theoretical and practical analysis.. Let us assume that we have a low pass signal modulated on a high frequency carrier.. now let us assume AWGN in the wireless channel.. now this will distort all frequencies.. now at the demodulator side, we have a low pass filter. If we have an ideal low pass filter, we can recover the low pass message signal along with low frequency noise superimposed on it. Now it is our responsisbility to first check the PSD of noise in this channel. Thus we know that the noise has a power P. If we now transmit a signal having power much much greater than this noise power, it will have very little affect on the final recieved signal. For a layman you can understand like you are riding a bicycle and wind is blowing side ways.. now a small wind will not distort u much (approximation considered) while a heavy wind might make you fall.. here the power in u and bike is equivalent to signal power and power in wind is analogus to the noise power.. Hope this helps..

 

[Y.SAI SARASWATHI]

Thank you ,
But my actual doubt is that in theory we consider that noise as Gaussian whose power spectral density is constant,but in actual it is not always going to happen ,but we are receiving signals accurately using some advanced technologies like rake receiver......so on,even in those there should be some prefixed algorithms to approximate power spectral density and some other parameters ,can you explain me how it is happening with such an accuracy.

 

[weetabixharry]

For any signal, it is the frequency components that matters.

This is far too general a statement. Frequency-domain and time-domain representations contain exactly the same information (and we can convert precisely between the two whenever we want). What you think "matters" is completely subjective. An enormous amount of signal processing and analysis are done in the time domain.

If the wave is very much attenuated but still the power of actual wave's frequency component is more than the superimposed noise, we can still recover it

Even if the noise is a million times more powerful than the signal, it may be possible to recover the signal. Your statements are simply not correct.

One way to look at the practical importance of Power Spectral Density in a communications system would be to consider the fundamental link between bandwidth and channel capacity. The Shannon-Hartley theorem tells us that channel capacity (the theoretical upper limit on bit rate) is directly proportional to bandwidth. Therefore, if we want to send data quickly, it must be spread over a broad frequency band (given a fixed SNR). For me, this is the most fundamentally important role of Power Spectral Density in a communications system.

 

[u7karsh]

weetabixharry

I started correctly but while writing, i could not express everthing i wanted to say thus resulting in this ambiguous text.. i really apologize for the same

 

[Y.SAI SARASWATHI]

Ok, thank you what about post no. 5