Positioning unmanned aerial vehicles under stochastic altitude, demand and location constraints

Many technological tools and equipment developed contribute to make human life easier in many areas from production to health, from security to education. As an engineering application, Unmanned Aerial Vehicles (UAV) are one of these tools that emerged as a result of technological developments. Although military purposes come to mind first for the area of use of UAVs, they have also been used for civilian purposes today due to the developments in their design and decreasing costs. One of these civilian areas of use of UAVs is to provide the function of a base station. Especially in disasters and emergencies, people turn to mobile communication tools to access both their immediate surroundings and information, and this causes communication to come to a complete breakdown due to the lack of sufficient infrastructure to meet the demand of service providers or their infrastructures being damaged. In such a case, UAVs mounted with small cell base stations can provide a low-cost connection to the regions thanks to their ability to provide rapid deployment. Within the scope of this research, a solution proposal will be made regarding the coordinates at which UAVs that can be used as temporary base stations should be positioned in order to meet customer demand and prevent communication from being interrupted in disasters or emergencies. Today, this problem is examined by considering different aspects such as coverage area, battery constraint, weather conditions. In this thesis study, the problem is aimed to determine the three-dimensional location of Unmanned Aerial Vehicles (UAV BS) that can be used as temporary base stations under stochastic user demand and location constraints and the number of UAV BSs that will be needed for the service in a way that will serve the maximum number of users and the problem is addressed in the Industrial Engineering discipline. An approach has been presented for the solution of this optimization problem. The algorithms used as the starting point of the approach are specified in the method section, and it is aimed to benefit from heuristics-meta heuristics and machine learning in particular. The number of users covered is used as the performance indicator. The thesis study differs from the studies in the literature in terms of the system scenario and solution method. It is assumed that the users move according to a movement model, the positions of the users are calculated for the relevant time intervals in line with the determined model, the users are divided into subgroups according to their locations with a clustering approach, and the problem is solved in a way that a UAV BS will serve each of these groups.