Highly Dynamic and Flexible Spatio-Temporal Spectrum Management with AI-Driven O-RAN: A Multi-Granularity Marketplace Framework

TL;DR

This work tackles the inflexibility of conventional spectrum management by proposing a GenAI-enabled marketplace within the O-RAN architecture that supports multi-granularity spatio-temporal-spectral management and dynamic trading across RT, near-RT, and non-RT timescales. By integrating generation and representation GenAI capabilities with AI-driven traffic forecasting, spectrum estimation, and allocation, the framework enables disaggregated AI agents to forecast demand and trade spectrum via an authorized broker, guided by policy centers and zero-touch SMO orchestration. Key contributions include the multi-granularity marketplace design, the six-way integration of GenAI with O-RAN components (Non-RT/ Near-RT RICs, xApps/rApps/dApps, O-CU/O-DU/O-RU, O-Cloud, SMO), and a three-control-loop architecture that achieves dynamic allocation and trading with demonstrated financial gains in numerical experiments. The proposed framework holds practical significance for 6G+ networks, enabling automated, adaptive spectrum sharing across 3D space, time, and frequency, and paving the way for scalable, revenue-enhancing, low-latency wireless ecosystems.

Abstract

Current spectrum-sharing frameworks struggle with adaptability, often being either static or insufficiently dynamic. They primarily emphasize temporal sharing while overlooking spatial and spectral dimensions. We propose an adaptive, AI-driven spectrum-sharing framework within the O-RAN architecture, integrating discriminative and generative AI (GenAI) to forecast spectrum needs across multiple timescales and spatial granularities. A marketplace model, managed by an authorized spectrum broker, enables operators to trade spectrum dynamically, balancing static assignments with real-time trading. GenAI enhances traffic prediction, spectrum estimation, and allocation, optimizing utilization while reducing costs. This modular, flexible approach fosters operator collaboration, maximizing efficiency and revenue. A key research challenge is refining allocation granularity and spatio-temporal dynamics beyond existing models.

Figures (7)

• Figure 1: A High-level Perspective of Our Proposed Dynamic and Flexible Marketplace-Based Spectrum-Sharing Framework.
• Figure 2: GenAI-Empowered O-RAN for Spectrum Sharing through Marketplaces.
• Figure 3: Actual and forecasted resource demand.
• Figure 4: Price of selling and buying SRUs corresponding to (a) one-hour and (b) one-minute time granularities.
• Figure 5: Reward values over training episodes.