This paper presents the distribution function of multi paths by use of two stochastic processes. The first process which is a Poisson process presents the paths occurring, and the second one which is an exponential process presents the paths standing. If the maximal paths are infinite, the Poisson distribution can be concluded. The notable error exists between the Poisson model and the measurement result. So the Poisson model is modified depending on the fact that the arrival rate decreases and the standing time get longer with the increase of arrived paths. The numeric results verify the consistence between the modified model and measurement.