Hi all,


why do we need higher SNR when multipath fading is considered than in a normal AWGN channel? My understanding is that multipath fading is equivalent to delay spread. What is the connection between delay spread and white noise?


thanks,
Aaron

Thanks.


I only know roughly what a Rake receiver is.

Anyway I think i'm starting to see where my misunderstanding is. I think SNR is a very basic definition of the ratio of the received signal power to the noise power. However, the received signal power includes components which may add constructively or destructively resulting in a drop in the SNR. Does that sound right? If so, then I mistakenly believed that SNR already takes into account the vector addition of different components of the received signal. If I'm wrong then I guess I need to read up more about the Rake receiver.


thanks,
Aaron

Quote:


I understand that the linear distortion causes ISI. Can such ISI be overcome by an improvement in the SNR?


thanks,
Aaron

Hi pancho_hideboo,


Quote:


that's really what I'm trying to understand actually.

I was looking up a bit about Rayleigh multipath channel model, and apparently the channel can be modeled as a vector addition of two signals with different attenuation, phase shift, and time shift. So ignoring the time shift for now, the actual power which we receive is ultimately attenuated due to the possible phase difference of the two signals (in the Rayleigh model). My understanding is that the SNR does not include the effect of this attenuation. Therefore, when comparing the Rayleigh model to the AWGN model, a higher SNR is required for the same BER.

Am I on the right track?

BTW the main reason I'm asking is I'm trying to figure out the SNR requirement for the IEEE 802.15.4 standard. Several papers have quoted an SNR requirement of about 0 dB, but I don't think they took multipath fading into account. The IEEE 802.15.4 standard says the SNR required is about 5-6 dB while one or two papers which study the effect of multipath fading for the 802.15.4 standard show that the required SNR is much higher (around 14 dB). So are they all talking about the same definition of SNR? I'm inclined to believe that works which include multipath fading...


thanks,
Aaron

Hi pancho_hideboo,


I have looked into the concepts you mentioned as well as matched filter detection, but the mathematics is quite involved, so its not easy to get a deep understanding. Anyway I've included a diagram of how I guess multipath affects the SNR requirement in a receiver. Does it look right? Basically |N| is the allowable noise power and |S| is the signal power. The cross hatched region represents the region where the PER would fall below the required value (1% in this case). These are constellation diagrams (I know I haven't labelled the axes).


thanks,
Aaron

thanks for the response. I think 802.15.4 was designed so that Rake receivers would be unnecessary.