Hi All,
This is just a brief update to the simulation results I presented previously in
http://www.ieee802.org/3/cg/public/adhoc/ocuanachain_8023cg_29_0317.pdf
Those simulations looked at the impact of a sinusoidal interferer at various frequencies. The question was posed as to what impact frequency spreading in the interferer would have on the PHY’s performance, specifically an AM interferer
rather than simply a sinusoid. So I repeated my simulations on the 1000m channel with an AM interferer with modulation as specified in IEC 61000-4-6:2007; a modulating sinusoid with frequency 1kHz and modulation index = 0.8. The figures below compare the SNR
with (i) Sinusoidal Interferer (ii) With AM interferer:
Interfering Sinusoid Interfering AM signal
This is as I was expecting, the 1kHz modulating signal has a negligible impact on SNR. This can be understood by looking at a typical equalizer Frequency Response:
The notch it implements to remove a sinusoidal interferer (at 2.25MHz in above plot) is broad relative to the +/- 1kHz spread due to the 1kHz modulating signal. So it will attenuate the AM signal almost as well as a pure sinusoidal interferer.
The performance results I have shown to date show acceptable performance in the presence of sinusoidal and AM interferers, I would still caution though that this is the steady-state performance, I do still have concerns about how well the
PHY will cope with the sudden appearance of such an interferer, eg the interferer ramps from 0mVpkpk to 75mVpkpk in ~zero time, this will produce errors at the slicer and I am concerned about the impact of error propagation in the DFE.
Oisín.