Resumen de Resource allocation and management techniques for network slicing in wifi networks
Matías Mario Richart Gutiérrez
Network slicing has recently been proposed as one of the main enablers for 5G networks; it is bound to cope with the increasing and heterogeneous performance requirements of these systems. To ``slice'' a network is to partition a shared physical network into several self-contained logical pieces (slices) that can be tailored to offer different functional or performance requirements.
Moreover, a key characteristic of the slicing paradigm is to provide resource isolation as well as an efficient use of resources. In this context, a slice is envisioned as an end-to-end virtual network which permits that the infrastructure operators lease their resources to service providers (tenants) through the dynamic, and on-demand, deployment of slices. Tenants may have complete control over the slice functions and resources, and employ them to satisfy their client’s demands.
Recent works on slicing for Radio Access Networks (RANs) just focus on general architectures and frameworks for the management and instantiation of network slices avoiding details on how the slices are implemented and enforced in the wireless devices.
Even more, while some techniques for slice enforcement already exist, most of them concentrate on cellular technologies, ignoring WiFi networks. Despite of their growing relevance and ubiquity, there are not many works addressing the challenges that appear when trying to apply slicing techniques over WiFi networks.
In this scenario, this thesis contributes to the problem of slicing WiFi networks by proposing a solution to enforce and control slices in WiFi Access Points. The focus of this work is on a particular and complex variant of network slicing called QoS Slicing, in which slices have specific performance requirements.
The main thesis contributions are divided in three: (1) a detailed analysis of the network slicing problem in RANs in general and in WiFi in particular, as well as a study and definition of the QoS Slicing problem, (2) a resource allocation model and mechanism for Wifi devices, and (3) a QoS Slicing solution to enforce and control slices with performance requirements in WiFi Access Points.
Given the novelty of the slicing concept and the complexity of the problem, a detailed study of the slicing problem was performed providing a comprehensive definition of the slicing concept, as well as a classification of the slicing variants. It is also introduced the two main problems of slicing wireless resources: resource allocation and isolation. In the scope of those problems, this thesis contributes with a novel approach where the resource allocation problem is divided on two sub-tasks: Dynamic Resource Allocation, and Enforcement and Control.
As a previous step to the construction of a QoS Slicing solution, it is proposed a novel method of proportionally distributing resources in WiFi networks, by means of the airtime. The proposed mechanism (called ATERR) is based on considering the airtime as the wireless resource to be shared and allocated. An analytical model of the ATERR algorithm is also developed, which shed light on how such resources could be split and on the capacities and limitations of the proposal.
The validity of the proposed model is assessed by means of a simulation-based evaluation on the NS-3 framework.
Finally, regarding the QoS Slicing problem, it is considered two different performance requirements: a guaranteed minimum bit rate and a maximum allowable delay.
The resource allocation problem to the different slices is formulated as a stochastic optimization problem, where each slice's requirement of bit rate and delay is modeled as a constraint.
A solution to the aforementioned problem is devised using the Lyapunov drift optimization theory to obtain an approximate deterministic problem. With this solution, it is developed a novel queuing and scheduling algorithm which allows implementing the obtained solution in WiFi devices.
