Recently, Unmanned Aerial Vehicle Base-Stations(UAV-BSs) are gaining popularity to provide wireless communications for various purposes. In particular, UAV-BSs can be actively utilized for a large number of users in areas where base station installation is difficult, inefficient, or restricted in communication use such as in battlefield environments. In order to satisfy the real-time and low-latency of continuous communication service in the environments where ground users move, it is necessary to predict the user's mobility to maintain the communication service quality and to change the location and altitude of the UAV-BS in real time. In this paper, the horizontal position of the UAV-BS is determined by predicting the movement of the ground user based on the EKF(Extended Kalman Filter). In addition the 3D position of the UAV-BS is updated in real time through vertical maneuvering by determining the altitude to maximize energy efficiency. To this end, the EKF measurement model was proposed using the minimum received power required in the downlink path of the UAV-BS and the user, and the effectiveness of the proposed method was evaluated through numerical simulations.