Log-distance path loss model is formally expressed as:
where
This corresponds to the following non-logarithmic gain model:
where c 0 = d 0 γ 10 − L 0 / 10 {\textstyle c_{0}={d_{0}^{\gamma }}10^{-L_{0}/10}} is the average multiplicative gain at the reference distance d 0 {\displaystyle d_{0}} from the transmitter. This gain depends on factors such as carrier frequency, antenna heights and antenna gain, for example due to directional antennas; and F g = 10 − X g / 10 {\textstyle F_{\text{g}}=10^{-X_{\text{g}}/10}} is a stochastic process that reflects flat fading. In case of only slow fading (shadowing), it may have log-normal distribution with parameter σ {\displaystyle \sigma } dB. In case of only fast fading due to multipath propagation, its amplitude may have Rayleigh distribution or Ricean distribution. This can be convenient, because power is proportional to the square of amplitude. Squaring a Rayleigh-distributed random variable produces an exponentially distributed random variable. In many cases, exponential distributions are computationally convenient and allow direct closed-form calculations in many more situations than a Rayleigh (or even a Gaussian).
Empirical measurements of coefficients γ {\displaystyle \gamma } and σ {\displaystyle \sigma } in dB have shown the following values for a number of indoor wave propagation cases.4
"Log Distance Path Loss or Log Normal Shadowing Model". 30 September 2013. https://www.gaussianwaves.com/2013/09/log-distance-path-loss-or-log-normal-shadowing-model/ ↩
Julius Goldhirsh; Wolfhard J. Vogel. "11.4". Handbook of Propagation Effects for Vehicular and Personal Mobile Satellite Systems (PDF). http://vancouver.chapters.comsoc.org/files/2016/05/handbook.pdf ↩
Wireless communications principles and practices, T. S. Rappaport, 2002, Prentice-Hall ↩