Resource Aggregation and Intelligent Scheduling for AR Remote Assistance

Immersive Augmented Reality (AR) remote assistance faces critical network transmission challenges due to dynamic data, heterogeneous resources, and complex resource interdependencies. This paper analyzes these unique characteristics and key challenges in the context of future wireless systems. To address them, we propose AIRAS, an AI-based architecture for network resource aggregation and intelligent scheduling. AIRAS introduces a Virtual Aggregation Network (VAN) for unified resource management and leverages Deep Reinforcement Learning (DRL) for dynamic resource balancing, aiming to ensure stable and reliable AR transmission. Simulation results demonstrate that AIRAS significantly outperforms traditional methods. Specifically, under dynamic network changes, it reduces average transmission time by 10 ms, improves throughput by 12%, and decreases packet loss by 4%. In high-concurrency scenarios, it achieves a 9 ms reduction in transmission time, an 8% throughput gain, and a 3% decrease in packet loss.