Radar networks show an interesting potential for safety and comfortable applications such as short-range automotive monitoring system or indoor monitoring. This paper presents our novel estimation algorithm of a target position. Especially we evaluate the performance about estimation accuracy and resistance to ghost targets under multipath environment. In above applications, the robust estimation is needed because the receivers tend to output corrupted measurement data. The corrupted data are mostly generated by multipath, sensitivity of receivers. As a result of computer simulations, our algorithm has fine accuracy and robust detections compared with a popular trilateration algorithm.