System design and resource allocation for persistent UAV presence (PUP)

Abstract

As the capability of unmanned aerial vehicles (UAVs) increases, there is increased interest in exploiting UAVs for continuous and long-term missions. However, due to UAV fuel restrictions, it is not possible to con-duct long-term missions with a few UAVs. Rather, a system that includes multiple UAVs and automated re-charging stations is required to provide automatic and persistent service. We are concerned with a particular class of missions in which the system should provide an uninterrupted UAV presence to predefined target are-as. Such missions are common in local surveillance and monitoring tasks. In order to construct a system providing persistent presence, one key question is to determine the number of resources required - we call this a system design. Further, the assignment of resources to tasks – we call resource allocation. We are also inter-ested in a generalization of the problem in which each target region requires a different service level. In the first chapter, we consider system design and resource allocation for the task of providing persistent UAV presence to multiple target areas with a single recharging station. The objective is to design the system which provides persistent presence that satisfies the required presence time ratio (service level) for each target. A decision-free Petri net model is developed that exploits the periodic nature of the problem and enables one to determine the number of resource required for fixed UAV routes. Combining the Petri net results with a search over feasible routes for the UAVs, a non-linear program with integer decision variables is developed to formalize the problem. A distance based heuristic and hybrid GA are developed to search for solutions and resolve the computational complexity of the problem. As the result, the number of resources and mission paths are derived. In the second chapter, the perspective is expanded to allow for two types of service stations: refueling sta-tions and central maintenance stations. UAVs can travel further using the refueling stations. A decision-free Petri net model for persistency is developed for cyclic paths including multiple immobile targets and stations. From the Petri net model, we derive a closed-form function for the minimum number of resources in the per-sistent system. A mathematical model that has the objective function derived from the Petri net is developed. To resolve the computational issue, a genetic algorithm (GA) is used to solve the problem. As the result, the minimum number of resources required and the mission path are derived. In the third chapter, resource allocation in a system where there are uncertain targets is conducted. Analyz-ing the number of resources required for uncertain targets is necessary to construct a new PUP system or pre-pare for a new target. We calculate 1) the expected number of resources for a single target uniformly distribut-ed in a disk centered at the station and 2) the expected number of resources required when the target’s location is assumed distributed in an arbitrary disk of given radius. We also analyze the properties of the system with uncertain targets and conduct a numerical study. Through the study, it is compared to the result from previous chapters. It is observed that the number of resources required increases inversely as the service area is widened to maximum capacity.