Optimal and probabilistic resource and capability analysis for network slice as a service
- Network Slice as a Service is one of the key concepts of the fifth generation of mobile networks (5G). 5G supports new use cases, like the Internet of Things (IoT), massive Machine Type Communication (mMTC) and Ultra-Reliable and Low Latency Communication (URLLC) as well as significant improvements of the conventional Mobile Broadband (MBB) use case. In addition, safety and security critical use cases move into focus. These use cases involve diverging requirements, e.g. network reliability, latency and throughput. Network virtualization and end-to-end mobile network slicing are seen as key enablers to handle those differing requirements and providing mobile network services for the various 5G use cases and between different tenants. Network slices are isolated, virtualized, end-to-end networks optimized for specific use cases. But still they share a common physical network infrastructure. Through logical separation of the network slices on a common end-to-end mobile networkNetwork Slice as a Service is one of the key concepts of the fifth generation of mobile networks (5G). 5G supports new use cases, like the Internet of Things (IoT), massive Machine Type Communication (mMTC) and Ultra-Reliable and Low Latency Communication (URLLC) as well as significant improvements of the conventional Mobile Broadband (MBB) use case. In addition, safety and security critical use cases move into focus. These use cases involve diverging requirements, e.g. network reliability, latency and throughput. Network virtualization and end-to-end mobile network slicing are seen as key enablers to handle those differing requirements and providing mobile network services for the various 5G use cases and between different tenants. Network slices are isolated, virtualized, end-to-end networks optimized for specific use cases. But still they share a common physical network infrastructure. Through logical separation of the network slices on a common end-to-end mobile network infrastructure, an efficient usage of the underlying physical network infrastructure provided by multiple Mobile Service Providers (MSPs) in enabled.
Due to the dynamic lifecycle of network slices there is a strong demand for efficient algorithms for the so-called Network Slice Embedding (NSE) problem. Efficient and reliable resource provisioning for Network Slicing as a Service, requires resource allocation based on a mapping of virtual network slice elements on the serving physical mobile network infrastructure. In this thesis, first of all, a formal Network Slice Instance Admission (NSIA) process is presented, based on the 3GPP standardization. This process allows to give fast feedback to a network operator or tenant on the feasibility of embedding incoming Network Slice Instance Requests (NSI-Rs). In addition, corresponding services for NSIA and feasibility checking services are defined in the context of the ETSI ZSM Reference Architecture Framework. In the main part of this work, a mathematical model for solving the NSE Problem formalized as a standardized Linear Program (LP) is presented. The presented solution provides a nearly optimal embedding. This includes the optimal subset of Network Slice Instances (NSIs) to be selected for embedding, in terms of network slice revenue and costs, and the optimal allocation of associated network slice applications, functions, services and communication links on the 5G end-to-end mobile network infrastructure. It can be used to solve the online as well as the offline NSIA problem automatically in different variants.
In particular, low latency network slices require deployment of their services and applications, including Network Functions (NFs) close to the user, i.e., at the edge of the mobile network. Since the users of those services might be widely distributed and mobile, multiple instances of the same application are required to be available on numerous distributed edge clouds. A holistic approach for tackling the problem of NSE with edge computing is provided by our so-called Multiple Application Instantiation (MAI) variant of the NSE LP solution. It is capable of determining the optimal number of application instances and their optimal deployment locations on the edge clouds, even for multiple User Equipment (UE) connectivity scenarios. In addition to that multi-path, also referred to as path-splitting, scenarios with a latency sensitive objective function, which guarantees the optimal network utilization as well as minimum latency in the network slice communication, is included. Resource uncertainty, as well as reuse and overbooking of resources guaranteed by Service Level Agreements (SLAs) are discussed in this work. There is a consensus that over-provisioning of mobile communication bands is economically infeasible and certain risk of network overload is accepted for the majority of the 5G use cases. A probabilistic variant of the NSE problem with an uncertainty-aware objective function and a resource availability confidence analysis are presented.
The evaluation shows the advantages and the suitability of the different variants of the NSE formalization, as well as its scalability and computational limits in a practical implementation.…
