AI-ML Beam Prediction with Association for Untransmitted Beamforming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI-ML based beam prediction systems fail to identify un-transmitted beams at the base station when they have not been beamformed earlier in the cell, hindering effective beamforming and scheduling of user equipment.
Innovation Solution
A pre-trained machine learning model at the user equipment predicts beam-related and UE location-related information for un-transmitted beams, using measurement results and association information, and transmits this data to the base station for beamforming.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the UE predicts un-transmitted beams based on measurement results of transmitted beams, then beam prediction capability is improved, but the base station cannot identify the predicted beams when they have not been beamformed earlier in the cell
Solution Approach 1:
The patent introduces association information as an intermediary element that links transmitted beams to un-transmitted beams. This association information acts as a mediator that enables the base station to identify predicted beams by establishing a relationship between beam sets, resolving the information loss problem without sacrificing prediction accuracy
Solution Approach 2:
The patent performs preliminary beamforming of transmitted beams before predicting un-transmitted beams. By pre-transmitting and measuring transmitted beams, the system builds up association information that enables subsequent identification of predicted un-transmitted beams, solving the identification problem proactively
2Productivity
If association information is provided between set A and set B beams, then beam prediction effectiveness is improved, but the system complexity increases when set A beams are not beamformed earlier in the cell
Solution Approach 1:
The patent implements a dynamic approach where association information is established on-demand between transmitted and un-transmitted beams rather than requiring pre-established static associations. This dynamic association mechanism maintains prediction effectiveness while reducing system complexity by only creating associations when needed
Data Source
AI summary
The subject matter relates to the wireless communication system. The subject matter discloses a method and system for predicting un-transmitted beams. The method includes determining a plurality of input parameters for a pre-trained machine learning model based on a set of transmitted beams received by a user equipment (UE) from a base station (BS). Further, the method discloses receiving association information between the set of transmitted beams and a set of un-transmitted beams, to be predicted. Finally, the method includes predicting, using the pre-trained model, beam-related information and UE location-related information associated with the set of predicted un-transmitted beams based on the plurality of input parameters and the association information. Further, the method discloses transmitting the predicted beam information to the BS, to enable the BS to serve the UE by beamforming the one of the beams from the set of predicted beams towards the UE.