Abstract:
In addition to the efforts in the next-generation cellular networks and traditional network services, the demand for a novel set of services leveraged through smart devices and artificial intelligence (AI) techniques increases tremendously. Pervasive healthcare, online gaming, augmented reality, smart city and many other service types with various performance and functional requirements are supplied with data generated by end-user devices. In this highly dynamic environment, the legacy network infrastructure and operations remain incapable of satisfying the expectations of the users and require ments of the services, especially those demanding real-time interaction with ultra-low latency. Therefore, this thesis focuses on the task offloading operations in a multi-tier edge environment and network slicing optimization problems to enable service-oriented behavior and address the demands of both operators and end-users. In this direction, an extensive literature review is carried out, the requirements are determined, and we provide a formal optimization model for each problem definition. In order to address the scalability issues and finding good quality solutions in a short time, heuristic so lutions are proposed. Besides efforts in optimization purposes, two different solution proposals using programmable network paradigms are provided as short-term and long term for implementing the service-centric behavior. The short-term solution based on Software-defined Networking (SDN) is further evaluated by implementing a fall-risk assessment service with real sensory data. The proposed solutions are novel and pro vide comprehensive guidance for operators and service providers on implementing a service-centric behavior and optimizing the operations in multi-tier edge systems.
