A Modular DTaaS Architecture for Predictive Slice Management in 6G Systems

Abstract

The sixth generation (6G) of wireless networks will require fundamentally new orchestration paradigms to meet stringent requirements for ultra-low latency, high reliability, and pervasive intelligence. Network slicing emerges as a key enabler to support diverse services with customized quality-of-service (QoS) guarantees. However, dynamic and fine-grained slice management poses significant challenges in terms of real-time provisioning, SLA assurance, and cross-layer observability. In this paper, we propose a novel Digital Twin as a Service (DTaaS) framework that embeds per-slice digital twins (SDTs) into the orchestration loop. Each SDT maintains a synchronized, real-time representation of its slice, leveraging multi-domain telemetry and deep sequential models to predict traffic evolution and SLA risks. The framework introduces modular intelligence layers, programmable interfaces, and edge-embedded decision-making to enable proactive provisioning, adaptive scaling, and closed-loop SLA assurance. Mathematical formulations for fidelity measurement, predictive control, and optimization objectives are provided to ensure rigor and transparency. Evaluation results demonstrate that DTaaS significantly improves SLA compliance ratio, reduces resource over-provisioning, and lowers average SLA violation probability, offering a scalable and reliable orchestration approach for 6G networks.