Work

Beyond Wide Area Connectivity: OSAF Multi-Domain Collaboration Optimization for Large-Scale Deterministic Networks

Year: 2026
Type: article
Abstract: Emerging service scenarios, such as large-scale engineering computing, have driven the evolution of network service requirements, shifting from traditional connectivity to deterministic quality assurance across wide-area networks. These scenarios demand stable end-to-end transmission capabilities. Furthermore, they also rely on the collaborative support of multiple deterministic domains, which are composed of multi-technology systems. Moreover, satisfying differentiated quality metrics is required for open service delivery in a multi-technology integrated network environment, making the construction of a wide-area deterministic fundamental network an inevitable trend. While IEEE TSN and IETF DetNet can ensure determinism within a single domain, they have obvious shortcomings in cross-domain collaboration and multi-domain joint optimization. To address these issues, this paper proposes a large-scale deterministic network architecture beyond wide-area connectivity based on open service availability first (OSAF). By adopting strategies for cross-domain collaboration and multi-domain joint optimization, it enables open service scheduling and path optimization across heterogeneous multi-domains. Additionally, this paper presents a Transformer-based deep reinforcement learning algorithm (T-DRL), where OSAF optimizes routing and scheduling decisions with a delay-priority objective, thereby enhancing the deterministic quality of service (QoS) guarantee for end-to-end differentiated open services. Empirical evaluations confirm substantial improvements in end-to-end delay, scheduling success rate, and service guarantees (jitter and packet loss rate) for both Best-Effort (BE) and DetNet flows, alongside reliable support for deterministic open service delivery and effective adaptation to differentiated QoS requirements in large-scale network scenarios.
Language: English
Open Access status: hybrid
Estimated APC paid: $2,645
OpenAlex ID: W7204268222
FWCI:
Cites: 48
Cited by:
Related to: 10