SANDWICH: Towards an Offline, Differentiable, Fully-Trainable Wireless Neural Ray-Tracing Surrogate

2024-11-13

arXiv.org

AI Summary

Conventional ray tracing (RT) incurs prohibitive computational overhead for high-frequency wireless channel modeling in B5G, while existing online learning approaches rely on real-time environmental monitoring and lack GPU acceleration capability. Method: This paper proposes the first offline, end-to-end differentiable, and fully GPU-trainable neural ray tracing framework. It formulates ray path generation as a scene-aware sequential decision process, integrating differentiable ray casting, joint optimization of scene encoding and channel response modeling, and GPU-native tensorized ray–scene intersection computation. Contribution/Results: The method eliminates dependence on real-time environmental supervision and unifies learning of geometric optics, electromagnetic wave propagation, and channel characteristics. Experiments demonstrate a ray angular error of only 0.04 radians and a channel gain estimation within 0.5 dB of the optimal physics-based RT baseline—significantly outperforming state-of-the-art online learning methods.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Hardware-aware ML
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Wireless ray-tracing (RT) is emerging as a key tool for three-dimensional (3D) wireless channel modeling, driven by advances in graphical rendering. Current approaches struggle to accurately model beyond 5G (B5G) network signaling, which often operates at higher frequencies and is more susceptible to environmental conditions and changes. Existing online learning solutions require real-time environmental supervision during training, which is both costly and incompatible with GPU-based processing. In response, we propose a novel approach that redefines ray trajectory generation as a sequential decision-making problem, leveraging generative models to jointly learn the optical, physical, and signal properties within each designated environment. Our work introduces the Scene-Aware Neural Decision Wireless Channel Raytracing Hierarchy (SANDWICH), an innovative offline, fully differentiable approach that can be trained entirely on GPUs. SANDWICH offers superior performance compared to existing online learning methods, outperforms the baseline by 4e^-2 radian in RT accuracy, and only fades 0.5 dB away from toplined channel gain estimation.

Problem

• Offline wireless ray-tracing surrogate model
• Differentiable and fully-trainable on GPUs
• Improved accuracy for B5G network signaling

Innovation

• Offline differentiable ray-tracing
• Generative models for signal properties
• GPU-based training without real-time supervision