Consider a standard RF channel over which signals are being transmitted over a 10 MHz bandwidth at 1800 MHz (using OFDM).

  • How does the channel's intrinsic attenuation vary over frequency?
  • More precisely, the subcarriers OFDM would be communicating over would be doing so over different bands of frequencies within the 10 MHz bandwidth.Would the channel attenuate the subcarriers differently based on whether they lie in the lower/higher frequencies within the bandwidth.
  • Is 10 MHz too small a bandwidth to observe any significant variations?

I am not referring to the Doppler spectrum (multi-path propagation) which causes small scale frequency spread and is a property of the physical surroundings, but to the attenuation that results from the medium itself.


No, there would be no local attenuation over the 10MHz subcarriers at 1800 MHz of any significance if you exclude the multipath effects. The standard path loss equation holds and this would be a non-dispersive (linear phase) channel with minimal difference in attenuation across the 10 MHz band. However, due to multipath fading you can see significant changes depending on the delay spread of the signal due to the fading environment (frequency selective fading).

See the figures below to get a sense of the delay spreads at 1.8 GHz for different environments and its effects on fading.

delay spread

channel distortion

In the simple model of a direct path and a delayed path both of equal amplitude, you would have frequency nulls spaced at $f= 1/T$ where T is the delay in seconds and f is the frequency spacing in Hz. This gives you a sense if the delay spread involved will result in flat or frequency selective fading. For the three typical delay spreads listed relevant to your 10 MHz channel, the spacing is 5 MHz (for 0.2 us open areas), 2 MHz (for 0.5 us suburban) and 333.3 KHz (for 3 us urban). Of course the nulls won't be so evenly spaced as in the figure as multiple delay paths are involved, but this helps give a first order metric as to the frequency selectivity of the fading.

  • $\begingroup$ The last paragraph was a good addition, I hadn't thought of that. $\endgroup$ – Television Jul 5 '16 at 12:31
  • $\begingroup$ Good to hear. Also if all the paths are reflected with no dominant direct path; the distribution of the received amplitude tends to be Raleigh, and if there is a dominant direct path with lots of reflections, the distribution tends to be Ricean. $\endgroup$ – Dan Boschen Jul 5 '16 at 15:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.