Wireless network operators are dealing with complex problems when planning network operations, particularly in an underground environment. To permit an acceptable level of automation in the planning process, simulation and optimization tools are being developed on a large scale basis. In this paper, a genetic algorithm is utilized to statistically analyse the narrow band wave propagation in a confined area. In our case this confined area is a real experimental gold mine. Four statistical fading models, namely, lognormal, Rice, Rayleigh, Nakagami were retained for a comparison process. Models and techniques are applied to many locations in underground galleries. A number of performance evaluation criteria (MSE and Komogorov Smirov test) were applied to comparatively assess model efficiency. A Genetic algorithm was used to produce all the numerical results presented in this study. The narrow band received power in the mining galleries is proven to follow a lognormal statistical power.