As the internet and mobile device get universal, mobile positioning in wireless communication becomes an important application. The wireless signal is influenced by many factors. Fading effect and multipath propagation increase the signal propagation time and change the angle of the received signal. All of these reduce mobile positioning accuracy. In line-of-sight (LOS) system, a signal can be measured precisely. However, there are many obstacles between a mobile station (MS) and a base station (BS) in the real environment. It causes non-line-of-sight (NLOS) effect. NLOS is a major factor that causes errors in mobile positioning. Although NLOS influences the signal propagation, these effects can be reduced. To enhance positioning accuracy, we choose the BS with the least error to estimate the mobile position by calculating the geometric dilution of precision (WGDOP), which is a value to quantify positioning performance. A WGDOP small value represents high accuracy. Using WGDOP to select the best combination of BSs improves MS positioning accuracy. Because the calculation of WGDOP is complex, it is a great burden on a device and spends lots of time. To reduce the load of devices, an approximation method based on genetic algorithm (GA) is proposed to derive WGDOP value efficiently.