A Dynamic Allocation Mechanism for Network Slicing as-a-Service

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
School of Business | Master's thesis
Degree programme
In my thesis, I explore the design of a market mechanism to socially efficiently allocate resources for network slicing as-a-Service. Network slicing is a novel usage concept for the upcoming 5G network standard, allowing for isolated and customized virtual networks to operate upon a larger, physical 5G network. By providing network slices as-a-Service, where the users of the network slice do not own any of the underlying resources, a larger range of use cases can be catered to. My market mechanism is a novel amalgamation of existing mechanism design solutions from economics, and the nascent computer science literature into the technical aspects of network slicing and underlying network virtualization concepts. The existing literature in computer science is focused on the operative aspects of network slicing, while economics literature is incompatible with the unique problems network slicing poses as a market. In this thesis, I bring these two strands of literature together to create a functional allocation mechanism for the network slice market. I successfully create this market mechanism in my thesis, which is split into three phases. The first phase allows for bidder input into the network slices they bid for, overcoming a trade-off between market efficiency and tractability, making truthful valuation Bayes-Nash optimal. The second phase allocates resources to bidders based on a modified VCG mechanism that forms the multiple, non-identical resources of the market into packages that are based on bidder Quality of Service demands. Allocation is optimized to be socially efficient. The third phase re-allocates vacant resources of entitled network slices according to a Generalized Second-Price auction, while allowing for the return of resources to these entitled network slices without service interruption. As a whole, the mechanism is designed to optimize the allocation of resources as much as possible to those users that create the greatest value out of them, and successfully does so.
Thesis advisor
Liski, Matti
mechanism design, Bayes-Nash, network slicing, 5G, allocation, dynamic allocation