Changing the noise model in WiFi

[Cosmos]

I can see that there are several files that measure the SNR of the received signal (nist-error-rate-model and dsss-error-rate-model). However, I can't see where they are deriving the noise model from. To calculate the SNR, the signal power will simply be the Received Power (which can be obtained from the propagation model and other factors), but how has the noise power been described in NS3? What is the default noise model?

I can't seem to find the source files that describe it. Can anyone point me in the right direction?

[Rediet]

Hello,

I think that you'll find that information in the InterferenceHelper class (https://www.nsnam.org/docs/models/html/wifi-design.html#interferencehelper) knowing that the thermal noise is set through the RxNoiseFigure attribute of WifiPhy.

Rediet

[Cosmos]

Hi Rediet,

Thanks for your response. I have another question: In the file src/wifi/model/yans-wifi-phy.cc, the decision whether a packet is successfully received or not is determined by: (Line 545)

if (m_random->GetValue () > snrPer.per)
{
   //Packet Reception Successful
}
else
{
   //Failure
}
 

where m_random is of type UniformRandomVariable. I'm not really able to understand how this works. How can a value generated, as per the uniform distribution, model the threshold for SNR? Shouldn't the threshold be a pre-decided one (As per BER vs. SNR tables)?

[Tommaso Pecorella]

That's not a BER or a SNR, that's a Packet Error Rate (PER). The PER depends on how long the packet is, and it is calculated from the BER (that is derived from the SINR).

Once you ave the PER, you simply decide with a random uniform variable. E.g., if you have a 10% BER, you get a random number between 0 and 1 and if it is below 0.1... well, bad luck. You have a 90% probability to have your random number above 0.1, and 10% below, which is what you wanted.

Cheers,

T.