SLA Decomposition for Network Slicing: A Deep Neural Network Approach

TL;DR

The paper tackles end-to-end SLA decomposition in multi-domain network slices managed by a two-level architecture (orchestrator and domain controllers). It introduces neural-network based risk models that are constrained to be monotonic, enabling accurate per-domain acceptance probabilities even with small datasets, and demonstrates six approaches to enforce monotonicity. Through experiments on synthetic multi-domain data, the authors show that AWET often yields the best balance of accuracy and efficiency for SLA decomposition, with other monotonic methods offering trade-offs in training time and robustness. The work advances scalable, online-adaptive decomposition of E2E SLAs by leveraging historical admission data and monotonic NN priors to produce reliable per-domain SLOs.

Abstract

For a network slice that spans multiple technology and/or administrative domains, these domains must ensure that the slice's End-to-End (E2E) Service Level Agreement (SLA) is met. Thus, the E2E SLA should be decomposed to partial SLAs, assigned to each of these domains. Assuming a two level management architecture consisting of an E2E service orchestrator and local domain controllers, we consider that the former is only aware of historical data of the local controllers' responses to previous slice requests, and captures this knowledge in a risk model per domain. In this study, we propose the use of Neural Network (NN) based risk models, using such historical data, to decompose the E2E SLA. Specifically, we introduce models that incorporate monotonicity, applicable even in cases involving small datasets. An empirical study on a synthetic multidomain dataset demonstrates the efficiency of our approach.

Figures (3)

• Figure 1: Network slicing management and orchestration system
• Figure 2: Ground truth and learned risk models
• Figure 3: Performance of SLA decomposition.