A Multi-Slice Multi-Operator Dynamic Weighted Fair Queuing RAN Scheduling in a Game Theoretic Framework

In 5G networks, multiple services with different Quality of Service (QoS) constraints are supported such as services requiring high throughput, low latency, or dense coverage. In order to support these different types of services, slicing is a key mechanism for providing logical partitioning of the same physical infrastructure based on the network service demands. Radio resource allocation optimization remains an important aspect in the context of slicing especially when multiple virtual operators with diverse types of services come into play. In such a context, numerous factors should be considered when allocating radio resources such as the available capacity, the diverse resource needs of virtual operators (VO) and the monetary cost of resources allotted by the infrastructure provider. This paper portrays the resource allocation problem as a non-cooperative game then as a Stackelberg game in a multi-VO multi-slice network. In such a context, the infrastructure provider seeks to satisfy the VO minimal throughput requirement via Dynamic Weighted Fair Queuing (DWFQ) while guaranteeing its own monetary profit, whereas VOs strive to selfishly fulfill their throughput requirements in any slice. Through exhaustive simulations, adequate pricing and satisfactory resource allocation to each VO in each slice were obtained based on specific network inputs.