Learning-based visibility prediction for terahertz communications in 6G networks

2024-09-01

Computer Communications

AI Summary

To address frequent line-of-sight (LoS) link disruptions caused by dynamic occlusions in 6G terahertz communications, this paper proposes a learning-based real-time visibility prediction method that estimates the LoS existence probability between transmitter and receiver with minimal overhead. The method innovatively integrates spatiotemporal graph neural networks with ray-tracing geometric priors, incorporating a lightweight LSTM, geometry-aware feature encoding, and synthetic data augmentation—thereby eliminating reliance on dense channel measurements. Evaluated in an urban microcell scenario, the approach achieves 92.7% prediction accuracy, end-to-end latency under 8 ms, and reduces energy consumption by 96% compared to conventional scanning schemes. To the best of our knowledge, this is the first work to enable millisecond-level, low-complexity online non-line-of-sight (NLoS) state prediction.

Technology Category

Planning, Routing, and Scheduling: Optimization of Spatio-temporal SystemsKnowledge Representation and Reasoning: Geometric, Spatial, and Temporal ReasoningMachine Learning: Graph-based Machine Learning

Application Category

Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environmentsGraph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphsSemantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Problem

• 6G Networks
• Terahertz Communication
• Blockage Prediction

Innovation

• 6G Networks
• Terahertz Communication
• Line-of-Sight Prediction