MicroOpt: Model-driven Slice Resource Optimization in 5G and Beyond Networks
Muhammad Sulaiman, Mahdieh Ahmadi, Bo Sun, Niloy Saha, Mohammad A. Salahuddin, Raouf Boutaba, Aladdin Saleh
TL;DR
MicroOpt tackles the challenge of dynamic, end-to-end resource allocation for 5G network slices under SLA constraints. It introduces a differentiable QoS model that outputs Gaussian parameters via a DNN and leverages the reparameterization trick to enable gradient-based optimization within a primal-dual, Lagrangian framework. The approach demonstrates up to about a 21.9% reduction in resource usage compared with state-of-the-art baselines on a real 5G testbed with live traffic traces, while respecting QoS degradation thresholds. Ablation studies confirm the importance of treating QoS as a distribution to enable effective gradient-based optimization, and the work validates MicroOpt in both simulation and hardware testbed settings, showing practical potential for adaptive, SLA-compliant network slicing in 5G and beyond.
Abstract
A pivotal attribute of 5G networks is their capability to cater to diverse application requirements. This is achieved by creating logically isolated virtual networks, or slices, with distinct service level agreements (SLAs) tailored to specific use cases. However, efficiently allocating resources to maintain slice SLA is challenging due to varying traffic and QoS requirements. Traditional peak traffic-based resource allocation leads to over-provisioning, as actual traffic rarely peaks. Additionally, the complex relationship between resource allocation and QoS in end-to-end slices spanning different network segments makes conventional optimization techniques impractical. Existing approaches in this domain use network models or simulations and various optimization methods but struggle with optimality, tractability, and generalizability across different slice types. In this paper, we propose MicroOpt, a novel framework that leverages a differentiable neural network-based slice model with gradient descent for resource optimization and Lagrangian decomposition for QoS constraint satisfaction. We evaluate MicroOpt against two state-of-the-art approaches using an open-source 5G testbed with real-world traffic traces. Our results demonstrate up to 21.9% improvement in resource allocation compared to these approaches across various scenarios, including different QoS thresholds and dynamic slice traffic.