Adaptive Beam Switching Using Predicted CQI for Load Balancing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing beam management systems in telecommunications networks poorly distribute active communication devices over beams, leading to inefficient resource utilization and suboptimal user throughput, especially when switching serving beams, and rely heavily on communication device vendor-specific measurements which can vary in accuracy.

Innovation Solution

Implement a machine learning-based method for adaptive beam management that predicts channel quality indicators (CQI) for target beam candidates, identifies offload candidates, and offloads communication devices to optimize resource allocation and improve user experience by considering additional metrics beyond RSRP, using a Random Forest Regressor model and a Beam Report Table.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional beam management switches serving beams based on RSRP measurements, then beam switching can be performed, but active communication devices become poorly distributed over beams and user throughput deteriorates

Engineering Contradiction:

Improvebeam switching accuracyVSAvoiduser throughput

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent changes the decision parameter for beam switching from RSRP (Reference Signal Received Power) to CQI (Channel Quality Indicator). This parameter change allows the system to consider not only signal strength but also channel conditions and interference levels, leading to more accurate beam switching decisions that improve device distribution and maintain user throughput

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary CQI prediction for multiple target beam candidates before actually switching beams. By predicting CQI values in advance using machine learning models and evaluating them against a threshold, the system can identify the best target beam beforehand, ensuring smooth transitions and preventing degradation of user throughput during beam switching

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If machine learning model predicts CQI for multiple target beam candidates, then beam switching accuracy improves, but computational complexity increases

Engineering Contradiction:

ImproveCQI prediction accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies partial action by predicting CQI for only the most promising target beam candidates rather than all possible beams. The system identifies a limited set of candidate beams based on current beam relationships and spatial considerations, then applies the computationally intensive machine learning prediction only to these candidates, reducing overall computational complexity while maintaining prediction accuracy

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent introduces an intermediary filtering mechanism that selects which target beam candidates warrant CQI prediction. By using geometric relationships, spatial filters, and candidate selection criteria as intermediaries, the system reduces the number of beams that require full ML model evaluation, thereby reducing computational complexity while preserving measurement precision for the most relevant candidates

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If beam switching is performed frequently to balance load, then resource utilization improves, but system stability deteriorates due to excessive switching

Engineering Contradiction:

Improveresource utilizationVSAvoidbeam configuration stability

Core Design Contradiction:

ProductivityVSStability of the object's composition

Solution Approach 1:

The patent implements feedback mechanisms where the system continuously monitors CQI values, beam load distribution, and switching history. This feedback allows the system to adjust beam switching decisions dynamically, preventing excessive switching by learning from past switching patterns and current system state, thereby maintaining stability while improving resource utilization

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies preliminary anti-action by evaluating multiple target beam candidates and comparing their predicted CQI values against thresholds before executing beam switches. This preliminary evaluation prevents premature or unnecessary switching by ensuring that switching only occurs when a significantly better beam candidate is identified, thus maintaining system stability while still enabling load balancing

Inventive Principle:

Principle #9Preliminary anti-action

Data Source

PatentUS12556254B2Adaptive beam management in telecommunications network

Publication Date: 2026.02.17 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

AI summary

A computer-implemented method for adaptive beam management for load balancing communication devices over ports used for a channel state information reference signal, CSI-RS, in a telecommunications network. The method includes predicting from a machine learning model predicted channel quality indicator, CQI, values for target beam candidates for offloading one or more communication devices. The method includes identifying a target beam candidate having a highest CQI value from the predicted CQI values. The method includes deciding to offload the one or more communication devices from an offload beam candidate to the target beam candidate when a difference between the highest CQI value of the target beam candidate and a CQI value of the offload beam candidate is greater than or equal to a threshold value. The method includes initiating offloading of the one or more communication devices from the offload beam candidate to the target beam candidate.