Flight mission-based service orchestration

• Autor(a) principal: Ramos, Joaquim Manuel Santos
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/38431
• topic: Drones; Kubernetes; Missions; Orchestration; Services; UAVs

UAVs are going through a swift growth. Their roots are military, but with new regulations and becoming more accessible, their use is expanding to multiple fields. They come as tiny and cheaper alternatives to helicopters when it comes to media uses. They can go anywhere the weather conditions and their sizes allow. Their deployment is easy and immediate, as is the case of multi-rotor drones because they only require a place of the size of their footprint to land and take off. UAVs are gaining traction to provide emergency services or to support emergency infrastructures. In this scenario they are used to deploy a plenitude of services, such as video streaming, communications access points, sensor gathering. All these services need to be maintained as a mission progresses. There is, however, a separation between the management of these services, and the system that controls the drones and their mission, so the maintenance of the services is done through manual processes. The services are deployed manually while not being tied to the mission that a drone is executing. This lack of synchronization leads to more time being wasted prepping the services, with the deployments having to be done manually. Therefore, in order to automate this process, we aim to bridge these systems together. This dissertation focus on bridging the service management side with an already existing mission planning framework, that allows the control of a fleet of drones. The service management is made possible by creating a Kubernetes cluster where the nodes executing the services are the drones. This solution allows automatic deployment of services/applications as declared in a mission script, and migrates them when needed. Through the creation of an intermediary entity we were able to allow the control framework to interact with the Kubernetes cluster. We tested the performance and the cost of the system, while also analysing the behaviour of it during the various steps of the implemented functionalities. We conclude that it allows for the automatic deployment and migration of the services, without introducing downtime, and that the introduction of this system did not put any strain on the hardware resources used.