In a public mesh network, we propose a control method to move mesh access points (APs) to optimal locations in response to traffic condition changes and evaluate the effectiveness of the move. To adapt the dynamic movement of pedestrian terminals accommodated in the mesh network, we configure a mesh network with in-UAV APs and move the APs with the UAVs. In the calculation to find the optimal location that minimizes the AP's moving distance and maximizes the user traffic accommodation, it is not realistic to find the exact solution in brute force ways. Thus, we propose a heuristic method (DMT method) that iteratively applies a minimum spanning tree method to determine the connectivity of all terminals and a spring balance model to determine the optimal AP locations. The proposed method was proven effective with the evaluation results in which throughput was improved up to 5.21 times compared to that before the move, and was approximately 95% of the exact solution.