frequency selective channel

Hi all! I have a rather major doubt and i would be more than thankful if someone could clear it. I am doing OFDM for my final project. I have simulated frequency selective fading channel and frequency selective non fading channel with OFDM. My question is: in the fading environment i can understand the benefits of OFDM (each subcarrier 'sees' a flat fading channel), but when it comew to non fading frequency selective channel (channel with a fixed impulse response) what are the advantages of doing OFDM? And why is the performance worse of AWGN? This means that the signal suffers from some channel characteristics? Is it right? And if so, which are they? Could you please help me it's very urgunt and i am rather confused.

I will appreciate every answer.
Thank you in advance.

Doesn't anybody have an answer?
Thanx:wink:

Does OFDM perform worse when no fading ?

when no fading the performance is better than fading, both OFDM..I know that my results are right..I just wonder if there's any advantage in doing OFDM when no fading and what is the primal reason for which the curve in the freq selective OFDM is worse than AWGN OFDM?

Your results are logical, OFDM is better when the channel is AWGN only this is very logical because there is no deep fade frequencies. OFDM mitigates the effect of deep fades but it doesn't cancel its effect completely. BTW this issue was discussed too many times in this forum but I think it has no clear answer till now!

But you can think of it as a fixed modulation scheme and two channels with the Rayleigh channel having more constraints (a worse channel).

Ahmed Alaa said:

Your results are logical, OFDM is better when the channel is AWGN only this is very logical because there is no deep fade frequencies. OFDM mitigates the effect of deep fades but it doesn't cancel its effect completely. BTW this issue was discussed too many times in this forum but I think it has no clear answer till now!

But you can think of it as a fixed modulation scheme and two channels with the Rayleigh channel having more constraints (a worse channel).

Click to expand...

Does your answer concern no fading freq selevtive channel? Deep fade frequencies are irrelevant of fading as it is defined...I think i am confused.. I am asking about the no fading freq selective case.
Thank you for your interest

Opps, I mis-read your post. Sorry !

Ahmed Alaa said:

Opps, I mis-read your post. Sorry !

Click to expand...

It's OK. Do you have any answer for the non fading freq selective case?

no, never dealt with this.

What is frequency selective "non fading"?

Don't think it is a valid term... doesn't make any sense whatsoever...

---------- Post added at 04:43 ---------- Previous post was at 04:40 ----------

Look, your channel impulse response will always be like [-1 -2 -3 -4 -5 ...] dB, i.e. like taps of an FIR filter. So, unless TimeSym >> DelSpread, you will always have Freq. Sel. Fading

---------- Post added at 04:47 ---------- Previous post was at 04:43 ----------

anta said:

when no fading the performance is better than fading, both OFDM..I know that my results are right..I just wonder if there's any advantage in doing OFDM when no fading and what is the primal reason for which the curve in the freq selective OFDM is worse than AWGN OFDM?

Click to expand...

Well, our systems are based on FDM, so why not use Orthogonal FDM.

---------- Post added at 04:49 ---------- Previous post was at 04:47 ----------

Ahmed Alaa said:

.... BTW this issue was discussed too many times in this forum but I think it has no clear answer till now!
......

Click to expand...

Let us get it over it, once and for all

Communications_Engineer said:

[/COLOR]Look, your channel impulse response will always be like [-1 -2 -3 -4 -5 ...] dB, i.e. like taps of an FIR filter. So, unless TimeSym >> DelSpread, you will always have Freq. Sel. Fading

Click to expand...

the impulse response of the fixed channel i am using is h=[ 9 9 8 8 7 6 ... 2 1]. Does this mean that there's fading?Because the impulse response is fixed... Could you plz clear this to me in detail? thanx

yes, it means it will have fading and that too frequency selective. Moreover, the phase response of the channel will also be non-linear (phase distortion)

try this to verify, freqz(h,1,1024,T), for any T = sample time

anta said:

the impulse response of the fixed channel i am using is h=[ 9 9 8 8 7 6 ... 2 1]. Does this mean that there's fading?Because the impulse response is fixed...

Click to expand...

Yes, h=[ 9 9 8 8 7 6 ... 2 1] is fading channel.

Perhaps you meant time-variant versus time-invariant channel? When impulse response is fixed over time, it's called time-invariant channel. When impulse response changes over time, it's called time-variant channel.

Fading versus non-fading depends on number of h elements. If h elements is more than 1, it is fading channel. Otherwise, it is non-fading.

Advantage of doing OFDM in time-invariant frequency selective fading is that you only need one-tap equalizer, hence receiver design becomes very simple.

Performance in fading is worse than AWGN because receiver can not perfectly know h. System suffers estimation error. It is normal to get performance degradation in fading channel.

Even though you perfectly know h, if your equalizer is just simply do inverse operation to h and multiply it with received signal (which is called Zero Forcing Equalization), there is noise enhancement in deep fade frequencies. Noise effects become larger at those frequencies and degrade performance more.

Thank you all for your answers. I finally got it this way:Fading means that the channel is time variant. Frequency selectivity means that he impulse response of the channel consists of more than one δ(t) hence it's response is non flat. A fading channel can have a single δ(t) which can change over time..thus it's not freq selective..On the same terms a freq selective channel can be non fading if the more than one δ(t) don't change over time. Is this right?

anta said:

Fading means that the channel is time variant.

Click to expand...

Fading can be time variant or time invariant. Time variant versus time invariant depends on doppler spread (usually affected by mobility). Fading versus non fading depends on presence of reflectors.

anta said:

Frequency selectivity means that the impulse response of the channel consists of more than one δ(t) hence it's response is non flat.

Click to expand...

I think for "impulse response of the channel consists of more than one δ(t)" is nearer to definition of fading channel, not frequency selective.

Fading channel can be frequency selective fading or flat fading. It depends on coherence bandwidth (affected by multipath spread) and signal bandwidth. When signal bandwidth is wider than coherence bandwidth, it becomes frequency selective.

anta said:

A fading channel can have a single δ(t) which can change over time..thus it's not freq selective..

Click to expand...

Yes, I agree. Single δ(t) which can change over time is time variant flat channel, not frequency selective.

anta said:

On the same terms a freq selective channel can be non fading if the more than one δ(t) don't change over time.

Click to expand...

No, it is time invariant fading channel, can be frequency selective or flat fading, depends on coherence bandwidth and signal bandwidth.

Reference: Fading - Wikipedia, the free encyclopedia