Technical Report : Analytical Modeling and Improvement of Interference-Coupled RAN Slicing

TL;DR

This technical report presents a queueing theoretic model for analyzing interference-coupled multi-cell RAN slicing, introduces KPIs for slice isolation, and proposes an interference-aware channel allocation policy to enhance isolation.

Contribution

It develops an analytical queueing model for interference in multi-cell RAN slicing and proposes an interference-aware channel allocation policy for better slice isolation.

Findings

The model accurately predicts network performance metrics.

The interference-aware policy improves slice isolation KPIs.

Numerical results validate the model and policy effectiveness.

Abstract

The emerging 5G technology needs to support simultaneously running incompatible service types on a common infrastructure. Network slicing is a solution that corresponds a slice of the network to each service type. Ensuring that user activity in one slice does not affect other slices, i.e., inter-slice isolation, is a key requirement of slicing. Since due to interference and channel conditions, wireless link quality is unpredictable and variable, providing isolation in radio access network (RAN) is cumbersome. In this technical report, we consider multi-cell RAN slicing where the coupled interference between cells results in dynamic behavior for slices. We propose a queueing theoretic-based model to analyze interference-coupled multi-cell RAN slicing. To this end, we map our scenario on a suitable state-dependent queueing network and propose an iterative algorithm to obtain approximately the network steady-state probability distribution and derive average delay and throughput. To quantify isolation in slices, we define some new key performance indicators (KPIs). Finally, we propose and analyze an interference-aware channel allocation policy that avoids use of overlapped frequency channels for as much as possible. Numerical results demonstrate the accuracy of our proposed model and the efficacy of the interference-aware policy in improving isolation-based KPIs compared to random allocation policy.