Control-Compute Governance in Agentic AI-RAN: AI Agents as Both Controllers and Workloads

2026-10-03

AI Summary

This study addresses the bidirectional control-computation coupling arising from AI agents sharing computational resources with the physical layer, which induces resource contention and control latency. We propose the first control-computation contract protocol bridging AI inference workloads and wireless control execution. Built upon the O-RAN architecture and large language model inference, this protocol coordinates AI-RAN workload governance with O-RAN control execution to achieve dynamic computational optimization. Over-the-air testbed evaluations demonstrate that the proposed protocol reduces the PHY timing budget violation rate from 20.7% to 0.9%, while the gNB control responsiveness significantly outperforms fixed-reservation baselines. This work establishes a standardized collaborative paradigm for AI-native wireless networks.

Abstract

Recent advances in artificial intelligence-radio access network (AI-RAN) are placing large language model (LLM)-driven agents within the open RAN (O-RAN) control hierarchy. A promising deployment for this agentic AI-RAN co-locates LLM inference with physical-layer (PHY) communication processing over the same accelerated compute pool for infrastructure sharing, data locality, and low control latency. However, this co-location induces bidirectional control-compute coupling, as the agent competes with PHY processing for compute, while its reasoned gNB control actions may alter future PHY workload and hence the compute available to its next inference. To this end, this article proposes the control-compute contract, a coordination protocol linking AI-RAN workload governance with O-RAN control execution. Specifically, we first expose direct compute contention on an over-the-air (OTA) O-RAN testbed, where continuous LLM inference increases the PHY decoding time approximately eightfold. We then formulate the contract as three protocol rules with operator-requirement protection, and map it onto O-RAN and AI-RAN functions as a five-stage workflow. Afterward, an OTA-calibrated case study of simulated uplink gNB control indicates that the contract keeps the proportion of slots violating the PHY time budget below $0.9\%$, against $20.7\%$ under ungoverned co-location, while achieving faster gNB control than a fixed compute reservation benchmark. Finally, we identify several open issues and outlooks for its practical deployment.

Problem

• AI-RAN
• control-compute coupling
• resource contention
• O-RAN
• LLM inference

Innovation

• Agentic AI-RAN
• Control-Compute Contract
• LLM Inference
• O-RAN
• Bidirectional Coupling