5Guard: Isolation-aware End-to-End Slicing of 5G Networks

TL;DR

The paper tackles the challenge of end-to-end isolation-aware 5G network slicing by formulating the online 5G-INS problem and proposing 5Guard, an ensemble framework that runs multiple optimization strategies in parallel to maximize InP profit under isolation overheads. It introduces three isolation levels ($L2$, $L1$, $L0$) and a detailed MIP formulation that captures NF/RU/VP mappings, capacity constraints, and QoS violation costs, then relaxes it to a tractable LP for scalable solving. The key novelty lies in coupling online SR provisioning with flexible isolation, revisitation/migration strategies, and a diverse algorithm ensemble that balances optimality and runtime deadlines. Empirically, 5Guard achieves up to 25.4% higher admission rates and up to 25.4% higher profits in large-scale networks compared with the best single algorithm, while revealing the trade-offs between isolation overheads and resource utilization, offering practical guidance for InPs deploying isolation-aware NS.

Abstract

Network slicing logically partitions the 5G infrastructure to cater to diverse verticals with varying requirements. However, resource sharing exposes the slices to threats and performance degradation, making slice isolation essential. Fully isolating slices is resource-prohibitive, prompting the need for isolation-aware network slicing, where each slice is assigned a tailored isolation level to balance security, usability, and overhead. This paper investigates end-to-end 5G network slicing with resource isolation from the perspective of the infrastructure provider, ensuring compliance with the customers' service-level agreements. We formulate the online 5G isolation-aware network slicing (5G-INS) as a mixed-integer programming problem, modeling realistic slice isolation levels and integrating slice prioritization. To solve 5G-INS, we propose 5Guard, a novel adaptive framework that leverages an ensemble of custom optimization algorithms to achieve the best solution within resource budget and time constraints. Our results show that 5Guard increases profit by up to 10.1% in resource-constrained environments and up to 25.4% in a real-world large-scale network compared to the best-performing individual algorithm. Furthermore, we analyze the trade-offs between isolation levels, their impact on resource utilization, and the effects of slice placement, demonstrating significant advantages over baseline approaches that enforce uniform isolation policies.

Figures (10)

• Figure 1: The proposed isolation-aware 5G network slicing system model. On top, an E2E 5G slice spanning from the RAN to the 5GC is illustrated, with slice components—NFs, RUs, and VPs—deployable on resources with three isolation levels: L2, L1, and L0. The left-hand side demonstrates three users of 5G slices: L2 for government and medical services, L1 for 5G-V2X services, and L0 for public streaming and entertainment services. Notably, the L2 slice is fully isolated, sharing no resources with other slices. In the example shown, servers' revisitation increases resource efficiency.
• Figure 2: The proposed isolation levels and resource overheads: For radio resources, these include dedicated NR-BS, PRBs for the entire radio frame with guard bands, and PRBs allocated via OFDMA. Transport isolation uses dedicated fiber, subcarriers with WDM/EON, and shared subcarriers with tagging. MEC/cloud employs servers, VMs, and containerized K8s pods.
• Figure 3: Flow diagram of the 5Guard framework for a 5G-INS instance. The ensemble algorithms, e.g., IAR, DTR, RNR, and OPT, operate concurrently on separate threads, processing shared input data. If the resource allocation deadline is reached, all algorithm executions are halted, and the best feasible solution is selected.
• Figure 4: 5G infrastructure topologies used in the evaluation
• Figure 5: The average SR's isolation level fed to isolation-aware approaches, the approach with no isolation, and complete isolation at each time step, with the shaded area showing the range of absolute isolation levels across the iterations

Theorems & Definitions (6)

• Theorem 1
• proof
• Theorem 2
• proof
• Theorem 3
• proof