Cognitive management of multi-service multi-tenant 5G mobile networks

Abstract:
Fifth generation (5G) mobile networks are expected to support diverse use cases beyond traditional voice and mobile broadband services. Efforts to connect anyone and anything have gained significant traction in several industries, among them automotive, logistics, manufacturing, and robotics. Exemplary use cases include eHealth, IoT (Internet of Things) applications, and V2X (vehicle-to-anything). Therefore, 3GPP LTE/EPC will have to evolve towards a flexible mobile network accommodating novel architectural principles, while at the same time maintaining backward compatibility. More specifically, in order to avoid deploying multiple physical networks addressing the respective use case, a single infrastructure needs to host multiple, logically separated dedicated networks, also referred to as network slices. This imposes new challenges on network management. While the objective is to manage each of the logical networks in a mostly isolated manner, the shared use of the underlying infrastructure will make “cross-slice” management functions mandatory. Further, the novel 5G mobile network architecture shall support legacy radio access technologies as well as novel radio interfaces, e.g., based on mm-wave or cm-wave transmissions. It should accommodate emerging processing paradigms such as mobile edge computing (MEC) and Cloud RAN (radio access network), while enabling flexible deployment patterns based on small, micro, and macro cells and allowing programmability to support very different requirements in terms of latency, robustness, reliability, and throughput. Cognitive network management is an important component to make such complex systems operable.