SANet: A Semantic-aware Agentic AI Networking Framework for Cross-layer Optimization in 6G

TL;DR

The paper introduces SANet, a semantic-aware AgentNet framework for cross-layer optimization in 6G that automatically infers user semantic goals and assigns cross-layer agents to fulfill them. A MoPS-based collaborative learning approach partitions large models into shared and agent-specific parts coordinated by embedding and gradient interfaces, with static- and dynamic-weighting algorithms delivering Pareto-optimal tradeoffs among competing agent objectives. The authors provide theoretical bounds on optimization, generalization, and conflict errors and validate the framework on a hardware RAN/core prototype, achieving up to 14.61% gains with 44.37% FLOPs and substantial reductions in training conflict. The work demonstrates a scalable, decentralized architecture for semantic-driven wireless optimization, offering new metrics, interfaces, and propagation of learned representations across layers. This could significantly improve adaptability and efficiency of future 6G networks via autonomous, goal-driven agent collaborations.

Abstract

Agentic AI networking (AgentNet) is a novel AI-native networking paradigm in which a large number of specialized AI agents collaborate to perform autonomous decision-making, dynamic environmental adaptation, and complex missions. It has the potential to facilitate real-time network management and optimization functions, including self-configuration, self-optimization, and self-adaptation across diverse and complex environments. This paper proposes SANet, a novel semantic-aware AgentNet architecture for wireless networks that can infer the semantic goal of the user and automatically assign agents associated with different layers of the network to fulfill the inferred goal. Motivated by the fact that AgentNet is a decentralized framework in which collaborating agents may generally have different and even conflicting objectives, we formulate the decentralized optimization of SANet as a multi-agent multi-objective problem, and focus on finding the Pareto-optimal solution for agents with distinct and potentially conflicting objectives. We propose three novel metrics for evaluating SANet. Furthermore, we develop a model partition and sharing (MoPS) framework in which large models, e.g., deep learning models, of different agents can be partitioned into shared and agent-specific parts that are jointly constructed and deployed according to agents' local computational resources. Two decentralized optimization algorithms are proposed. We derive theoretical bounds and prove that there exists a three-way tradeoff among optimization, generalization, and conflicting errors. We develop an open-source RAN and core network-based hardware prototype that implements agents to interact with three different layers of the network. Experimental results show that the proposed framework achieved performance gains of up to 14.61% while requiring only 44.37% of FLOPs required by state-of-the-art algorithms.

Figures (52)

• Figure 2: (a) System model, and (b) operational workflow for a general AgentNet-based cross-layer optimization framework for wireless networks.
• Figure 3: Model training procedures of (a) static-weighting and (b) dynamic-weighting algorithms, and (c) inference procedures of both algorithms.
• Figure 4: SANet prototype.
• Figure : (a) aAgent
• Figure : (a)