Neural Channel Knowledge Map Assisted Scheduling Optimization of Active IRSs in Multi-User Systems

2025-08-09

AI Summary

To address severe double path loss, high channel-state-information (CSI) acquisition overhead, and excessive scheduling complexity in multi-intelligent reflecting surface (IRS) multi-user systems, this paper proposes a lightweight resource scheduling framework based on a neural Channel Knowledge Map (CKM). We innovatively design two cascaded Transformer networks—LPS-Net and SE-Net—to jointly model historical CSI and throughput data, enabling accurate prediction of link power and ergodic spectral efficiency. Furthermore, we propose a low-complexity hybrid scheduling algorithm, Stable Matching–Iterative Balancing (SM-IB), which approximates the max-min throughput optimal solution in polynomial time. Experiments demonstrate that the proposed method reduces scheduling complexity by O(2^N) compared to exhaustive search, decreases channel prediction error by 37.2%, and improves throughput fairness by 28.5%, thereby effectively supporting dense-user deployments of proactive IRS systems.

Technology Category

• Planning, Routing, and Scheduling: Scheduling
• Intelligent Robots: Multi-Robot Systems
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems

Abstract

Intelligent Reflecting Surfaces (IRSs) have potential for significant performance gains in next-generation wireless networks but face key challenges, notably severe double-pathloss and complex multi-user scheduling due to hardware constraints. Active IRSs partially address pathloss but still require efficient scheduling in cell-level multi-IRS multi-user systems, whereby the overhead/delay of channel state acquisition and the scheduling complexity both rise dramatically as the user density and channel dimensions increase. Motivated by these challenges, this paper proposes a novel scheduling framework based on neural Channel Knowledge Map (CKM), designing Transformer-based deep neural networks (DNNs) to predict ergodic spectral efficiency (SE) from historical channel/throughput measurements tagged with user positions. Specifically, two cascaded networks, LPS-Net and SE-Net, are designed to predict link power statistics (LPS) and ergodic SE accurately. We further propose a low-complexity Stable Matching-Iterative Balancing (SM-IB) scheduling algorithm. Numerical evaluations verify that the proposed neural CKM significantly enhances prediction accuracy and computational efficiency, while the SM-IB algorithm effectively achieves near-optimal max-min throughput with greatly reduced complexity.

Problem

• Optimize active IRS scheduling in multi-user systems
• Reduce channel state acquisition overhead and delay
• Achieve near-optimal throughput with low complexity

Innovation

• Transformer-based DNNs predict spectral efficiency
• Neural Channel Knowledge Map enhances prediction
• Low-complexity SM-IB algorithm optimizes scheduling