Wireless Spatial Filter Prediction for NR Beam Sweeping

The patent applies preliminary action by pre-training a neural network model offline with comprehensive beam measurement data. The trained model is then deployed on terminal devices to perform rapid beam prediction without requiring exhaustive beam sweeping at runtime. This resolves the contradiction by performing the computationally intensive work in advance, enabling fast real-time beam selection with reduced delay while maintaining accuracy through the pre-trained model's predictions.

The patent uses copying by creating a virtual model (neural network) that replicates the complex beam selection process. Instead of physically traversing all beam combinations, the system copies the essential patterns and relationships learned from comprehensive measurements into the neural network model. This virtual copy enables rapid prediction of optimal beams without repeating the exhaustive physical sweeping process, thereby reducing overhead and delay while preserving selection accuracy.

The patent applies preliminary action by performing comprehensive beam measurements and training the neural network model offline before actual operation. During runtime, the pre-trained model predicts optimal beams without requiring the terminal to sweep through all possible beam combinations. This resolves the contradiction by completing the exhaustive exploration phase in advance, enabling the system to achieve complete beam coverage understanding while reducing real-time signaling overhead to only the necessary beam prediction results.

The patent uses copying by creating a neural network model that captures the complete beam space characteristics and relationships. This virtual model copies the essential information about all possible beam combinations and their performance characteristics, allowing the system to achieve complete beam coverage knowledge without repeatedly transmitting exhaustive beam measurement data. The model serves as a compact representation that reduces overhead while maintaining adaptability across different beam scenarios.