Downlink Beam Prediction Model Selection Under Channel Coherence Time
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack a method to determine an appropriate beam prediction model for downlink beam management, particularly in selecting parameters K and F for AI/ML models, which affects prediction accuracy.
Innovation Solution
Determine a constraint condition for selecting a prediction model based on the coherence time of the channel, ensuring that the time difference between the earliest historical beam measurement and the latest future prediction information is within the coherence time, thereby facilitating the selection of a more accurate prediction model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI/ML model is used for beam prediction, then beam management overhead is reduced, but prediction accuracy deteriorates due to lack of constraint conditions
Solution Approach 1:
The patent applies parameter changes by introducing constraint conditions based on coherence time and time differences. Specifically, it constrains the time difference between the earliest historical beam measurement and the latest future prediction information to be within the coherence time of the channel. This parameter-based constraint ensures that the AI/ML model operates within valid temporal boundaries, maintaining prediction accuracy while achieving efficient beam management.
2Adaptability or versatility
If historical beam measurement data is used for training, then prediction model is established, but channel environment applicability deteriorates when time exceeds coherence time
Solution Approach 1:
The patent applies preliminary action by pre-determining the coherence time of the channel and establishing constraint conditions before performing beam prediction. The system calculates the coherence time based on channel characteristics and uses this pre-established parameter to constrain the time difference between historical measurements and future predictions. This preliminary constraint ensures that the training data and prediction results remain applicable to the current channel environment, maintaining reliability.
Data Source
AI summary
The electronic device of present invention comprises a processing circuit configured to: determine, on the basis of the movement speed of a terminal device, a coherence time of a channel where the terminal device is located; and determine, on the basis of the coherence time, a restriction condition for selecting, for the terminal device, a prediction model for downlink beam prediction. The prediction model determines successive F pieces of prediction information for a downlink beam based on K pieces of measurement information obtained by successive K times of beam measurement for the downlink beam on a terminal device side, wherein K and F are each a natural number. The restriction condition comprises: a time from first beam measurement in the K times of beam measurements to the last piece of prediction information in the F pieces of prediction information is less than or equal to the coherence time.