3GPP shadow fading in 5G: log-normal or normal-random

[mjaew1]

Dear colleagues,

In 3GPP TS 38.901 (chapter 7.4.1) the distribution of shadow fading is stated to be log-normal.

When I was studying the GetShadowing function* in three-gpp-propagation-loss-model.cc I see the following:

if (notFound || newCondition)

{

// generate a new independent realization

shadowingValue = m_normRandomVariable->GetValue () * GetShadowingStd (a, b, cond);

}

else

{

// compute a new correlated shadowing loss

Vector2D displacement (newDistance.x - it->second.m_distance.x, newDistance.y - it->second.m_distance.y);

double R = exp (-1 * displacement.GetLength () / GetShadowingCorrelationDistance (cond));

shadowingValue = R * it->second.m_shadowing + sqrt (1 - R * R) * m_normRandomVariable->GetValue () * GetShadowingStd (a, b, cond);

}

Using normRandomVariable made me wonder why that was chosen, when 38.901 states lognormal. Essentially this is my question.

I did see there is another class called LogNormalRandomVariable and in there we have the function GetValue** which returns a random unsigned integer from a log-normal distribution based on the provided mean and std dev. However that is not used in the DoCalcRxPower for ThreeGppUmaPropagationLossModel. Instead, the normRandomVariable is used.

I am new to understanding the model built here so may have misinterpreted the code (hopefully not). Any help appreciated.

Kind regards,

Matthew.

* line 186 here: https://www.nsnam.org/doxygen/three-gpp-propagation-loss-model_8cc_source.html)

** line 831 of https://www.nsnam.org/doxygen/random-variable-stream_8cc_source.html