DL-RS Beam Pair Prediction Using ML Training Data Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing beam management systems in wireless communication, particularly in high-frequency range (FR2), face challenges in efficiently predicting beam pair links due to the lack of knowledge about wireless device panels and beams, leading to prolonged time in determining suitable wireless device beams, which can significantly reduce performance when devices are moving or rotating.
Innovation Solution
A method involving a trained machine learning model at the network node for beam pair link prediction, where a wireless device assists in data collection by receiving configurations, performing measurements, and providing assistance information to map network node beams with wireless device beams, thereby facilitating efficient beam pair link prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional beam management procedures are used to determine suitable wireless device beams, then beam pair links can be established, but the time required to determine suitable beams is prolonged, reducing performance when devices are moving or rotating
Solution Approach 1:
The system performs preliminary data collection during normal beam management operations, gathering measurement data from reference signals without interrupting ongoing communications. This pre-collected data is then used by the machine learning model to predict beam pair links in advance, eliminating the need for time-consuming beam searches when devices move or rotate.
Solution Approach 2:
The patent replaces traditional mechanical beam management procedures (systematic beam sweeping and measurement) with an intelligent prediction system based on machine learning. The ML model analyzes historical measurement data and device capability information to predict suitable beam pair links, substituting the mechanical search process with computational prediction.
2Measurement precision
If comprehensive beam measurements are performed to ensure accurate beam pair link determination, then prediction accuracy improves, but the complexity of the beam management system increases
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between raw measurement data and beam pair link determination. The ML model processes measurement data along with device capability information, transforming complex raw data into reliable predictions without requiring the network to directly manage all measurement details.
Solution Approach 2:
The system uses existing reference signals (CSI-RS, SSB) for multiple purposes: both for traditional channel state information acquisition and for collecting data to train the beam pair link prediction model. This multi-functional use of existing signals reduces the need for additional measurement overhead and system complexity.
3Reliability
If more beam data is collected for machine learning model training, then prediction accuracy improves, but the signaling overhead and processing requirements increase
Solution Approach 1:
The system enables wireless devices to self-report their capability information regarding beamforming and antenna configurations. Devices autonomously provide this information without requiring detailed network queries, reducing signaling overhead while ensuring the ML model receives necessary data for accurate predictions.
Solution Approach 2:
The patent combines multiple types of information (measurement data from reference signals, device capability information, and beam configuration data) into a unified training dataset for the machine learning model. This consolidation approach maximizes the value of collected data while minimizing redundant signaling and processing requirements.
Data Source
AI summary
A method, system and apparatus are disclosed. In one or more embodiments, a wireless device is configured to receive a beam pair link configuration indicating a plurality of beam pair links, where each beam pair link corresponds to a mapping between a network node beam and a wireless device beam. The wireless device is further configured to perform at least one measurement of at least one of the plurality of beam pair links.