LOCAL RECONFIGURATION OF 5G NETWORK SLICES UNDER NODE FAILURES AND OVERLOADS

Анотація

Relevance. The rapid deployment of 5G networks and the widespread use of network slicing create new challenges for ensuring service reliability and resilience. Virtualized infrastructures based on Network Functions Virtualization increase flexibility but also introduce higher failure rates and performance variability. In such environments, centralized recovery mechanisms often fail to meet strict latency requirements, especially for ultra-reliable low-latency communication services. Research object: The research object is the process of failure recovery and resource reconfiguration in virtualized 5G network slicing environments under node failures and progressive overload conditions. Purpose: The purpose of the study is to develop an efficient distributed method for local reconfiguration of network slices that enables rapid recovery of virtual network functions while considering slice priorities, migration costs, and latency constraints. Results. A distributed local reconfiguration (DLR) framework is proposed, based on a hierarchical architecture consisting of a global orchestrator, regional slice managers, and local monitoring agents. The approach introduces a multi-objective optimization model for slice manager placement and a unified migration cost function that considers computational, network, disruption, and SLA penalty components. Localized recovery algorithms are developed to handle both catastrophic node failures and progressive overload scenarios while prioritizing slices according to their service requirements. Conclusions. The proposed distributed slice recovery framework enables fast and scalable reconfiguration of 5G network slices under failure conditions. By combining slice-aware prioritization, cost-aware migration decisions, and localized management, the approach improves recovery speed and operational efficiency while preserving service quality for latency-critical applications.