This paper deals with the characterization of proposed modified random direction (M-RD) mobility model in unmanned aerial vehicles (UAVs) deployed as base stations for serving terrestrial users within finite cellular network. This mobility model is the upgradation to classical random direction (RD) mobility model with inclusion of several stops within the path to improve the probability of serving users. The simulation results of proposed M-RD mobility model in terms of distribution, density and average rate are compared with prevalent mobility models, i.e., random walk (RW) and random waypoint (RWP). Distribution plots showing positive skewness for RW/RWP and negative skewness for M-RD, reveal a framework for obtaining uniformity in the distribution with an adjustment of proportion of UBSs following RWP and M-RD. The density of the network ofinterfering UBSs for M-RD exhibits better homogeneity at large time intervals. Also, the average rate achieved by a typical user located at different locations in the environment has shown remarkable improvement in M-RD as compared to RW and RWP, especially at large time intervals.