Agentic AI-Driven Decision Support for 6G Networks: An Explainable Fuzzy Multi-Criteria Architecture for Autonomous Service Assurance

Abstract

The transition toward sixth generation (6G) communication systems necessitates a paradigm shift from connectivity-centric architectures to intelligence-native infrastructures capable of autonomous, context-aware decision-making. This study proposes a Layered Agentic Core Decision Support Architecture (LAC-DSA) for 6G networks, wherein Large Language Models (LLMs) and autonomous AI agents collaborate to optimize network functions—including resource allocation, compute offloading, and service orchestration—while maintaining Service Level Agreement (SLA) compliance. The architecture is grounded in the emerging IMT-2030 requirements and draws conceptual inspiration from the Model Context Protocol (MCP) and Agent-to-Agent (A2A) interoperability frameworks. A Pythagorean Fuzzy CRITIC (CRiteria Importance Through Intercriteria Correlation)–TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) methodology is developed for explainable multi-criteria action ranking, integrating objective criterion weighting with distance-based compromise ranking under uncertainty. The architecture is analytically demonstrated through a smart vehicle ecosystem case study encompassing five use cases. The Pythagorean Fuzzy TOPSIS analysis identifies load-balanced edge offloading as the optimal action for latency-critical perception tasks, while a complementary FRAM (Functional Resonance Analysis Method)–FMEA (Failure Mode and Effects Analysis) security risk prioritization reveals token and personally identifiable information (PII) leakage and attribute-based access control (ABAC) misconfiguration as the highest-priority mitigation targets. Sensitivity analysis across varying decision-maker attitudes and weight perturbations confirms the robustness of the ranking outcomes. The study further maps eight proposed service requirements to 3GPP, ITU-R IMT-2030, ETSI ZSM, TM Forum Autonomous Networks, and O-RAN standardization frameworks, contributing to the foundation for normative 6G specifications.