Network Node AI Risk Thresholding for QoS Protection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The integration of AI modules in communication networks poses risks due to potential wrong decisions caused by inadequate training, abnormal data behavior, or unexpected events, leading to performance degradation and conflicts with quality of service.

Innovation Solution

A mechanism is introduced to quantify and manage the risk of using AI modules by comparing the level of risk with a threshold, deactivating the module when the risk exceeds the tolerable level, and retraining it when abnormal behavior is detected, using a Markov model to predict performance degradation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If AI modules are used to manage communication features, then network performance and efficiency are improved, but the risk of wrong decisions and performance degradation increases

Engineering Contradiction:
Improvenetwork performanceVSAvoiddecision accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the network node continuously monitors the actual performance of AI-managed features and compares it with expected performance. When performance degradation is detected, the system provides feedback to deactivate or retrain the AI module, creating a closed-loop control system that adapts to changing conditions and prevents wrong decisions from causing lasting harm.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent performs preliminary risk assessment and performance monitoring before AI modules can cause significant degradation. By continuously evaluating performance metrics and detecting abnormal behavior early, the system takes preventive action to deactivate or retrain AI modules before wrong decisions lead to severe performance degradation or QoS conflicts.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If AI modules are continuously monitored and retrained, then reliability is improved, but device complexity and processing requirements increase

Engineering Contradiction:
ImproveAI module safetyVSAvoidmonitoring mechanism
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements monitoring and risk assessment locally at each network node that deploys AI modules, rather than requiring centralized control. Each node independently evaluates its own AI module performance and makes local decisions about activation and deactivation, reducing the complexity of the overall monitoring system while maintaining high reliability.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The network node performs self-monitoring and self-assessment of AI module performance without requiring external intervention. The node autonomously detects abnormal behavior, evaluates risk levels, and decides when to deactivate or retrain AI modules, reducing the need for complex external monitoring infrastructure.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4331258B1Network node and method performed therein
Publication Date: 2026.01.07 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP4331258B1 patent drawingFigure 1
  • EP4331258B1 patent drawingFigure 2
  • EP4331258B1 patent drawingFigure 3a

AI summary

Embodiments herein relate to a method performed by a network node (12,13) for managing communication in a communication network. The network node (12,13) obtains an indication relating to a level of risk in using an AI module for managing a feature, wherein the level of risk is associated with a level of a degradation of a performance in the communication network. The network node compares the obtained level of risk with a set level of risk; and based on the comparison, activates or deactivates the AI module for managing the feature.