2023 article
LEONS: Multi-Domain Network Slicing Configuration and Orchestration for Satellite-Terrestrial Edge Computing Networks
ICC 2023 - IEEE INTERNATIONAL CONFERENCE ON COMMUNICATIONS, pp. 6294–6300.
author keywords: multi-domain network slicing; STECNs; energy cost; SLA; restless multi-armed bandit (RMAB)
TL;DR:
This paper presents a multi-domain network slicing scheme for satellite-terrestrial edge computing networks (STECNs) that admits different slice configurations and model the slice/resource availability as a Markov process to track the probability of achieving the SLA per configuration.
(via Semantic Scholar)
UN Sustainable Development Goal Categories
9. Industry, Innovation and Infrastructure
(Web of Science)
Source: Web Of Science
Added: March 4, 2024
In this paper, we present a multi-domain network slicing scheme for satellite-terrestrial edge computing networks (STECNs) that admits different slice configurations. Each slice is configured to include terrestrial-air, terrestrial-satellite, terrestrial-air-satellite, or terrestrial-air-satellite-gateway domain topologies. However, the multi-domain nature of STECNs makes slicing especially challenging since the cross-domain orchestrator has no knowledge of the resource availability in different domains. Our goal is to design an algorithm that builds a belief in resource availability to jointly optimize the slice configuration, service level agreement (SLA) decomposition, routing, and resource allocation. We model the slice/resource availability as a Markov process to track the probability of achieving the SLA per configuration. To solve the multi-domain slicing problem, the cross-domain orchestrator interacts with the configuration coordinator to define an index-based slice configuration policy based on restless multi-armed bandits (RMABs), which is aware of the network traffic. The configuration coordinator decomposes the SLA and each domain controller solves the optimum routing and resource allocation. Our slicing scheme is evaluated using five typical application scenarios for STECNs. Simulation results show that our scheme achieves six times higher reward than agnostic schemes and efficiently performs multi-domain slicing with low complexity.