Agentic AI for Bilevel Long-Term Optimization of Policy-Driven Physical Layer Systems

2026-06-23

AI Summary

Existing approaches struggle to adapt to the dynamic changes in network policies, service demands, and real-time constraints, thereby limiting long-term performance optimization of physical-layer systems. This work proposes Agentic-LTPO, a novel framework that introduces agentic AI into a two-tier long-term optimization architecture: an upper-level agent synthesizes policy directives, environmental conditions, and historical experience to generate adaptive configurations, while a lower-level module executes real-time physical-layer decisions based on these configurations. The framework innovatively incorporates a policy-driven adaptive configuration mechanism and a multi-agent collaborative decision-making approach enhanced by retrieval-augmented experience validation, integrated with a closed-form beamforming algorithm. Evaluated in a cell-free MIMO setting, the system achieves a 57.2% improvement in long-term performance over conventional methods, demonstrating significantly enhanced adaptability to dynamic policy shifts.

Abstract

Network operators' changing policies, service requirements, and stringent real-time constraints render existing methods designed with fixed objectives and constraints ineffective. This paper presents Agentic long-term performance optimization (Agentic-LTPO), a nested bilevel optimization framework that can be applied to adaptive physical layer problem configuration. The key idea is to employ agentic AI to generate upper-level configurations in a bilevel optimization structure, where evolving operator policies, environment summaries, and historical experiences are translated into structured lower-level optimization problem configurations. The lower level solves the problems with updated configurations for real-time physical-layer decisions. Considering cell-free MIMO beamforming as a use case, we embody Agentic-LTPO by designing a new multi-agent decision process with retrieval-augmented experience-based verification in the upper level, together with a closed-form beamformer in the lower level. Experiments demonstrate that Agentic-LTPO exhibits strong adaptability to dynamic operator policies and effectively enhances the system's long-term performance by 57.2% compared to traditional methods.

Problem

• bilevel optimization
• agentic AI
• policy-driven systems
• long-term optimization
• physical layer

Innovation

• Agentic AI
• bilevel optimization
• long-term performance optimization
• cell-free MIMO
• retrieval-augmented verification