Dynamic Network Configuration With Artificial-Life Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communications networks face challenges in configuring alarms that are either too sensitive, triggering false alarms, or not sensitive enough, leading to delayed issue detection and resolution, and current Self-Organizing Network (SON) features are implemented separately, requiring complex algorithms and manual tuning.
Innovation Solution
A method and apparatus using a population of machine-learning processes that operate based on diverse data models and decision-making rules to recommend configuration changes by producing individual recommendations, which are then aggregated to form an output recommendation through selection or clustering, and evolve to improve accuracy over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If alarm sensitivity is increased to detect issues earlier, then detection timeliness is improved, but false alarm rate increases reducing reliability
Solution Approach 1:
The alarm system is segmented into multiple independent machine-learning processes (e.g., detection module, analysis module, decision module) that operate with different data models and decision-making rules. Each segment processes alarm data independently and contributes to the final alarm determination, allowing the system to maintain high sensitivity while filtering false alarms through collective decision-making.
Solution Approach 2:
The system dynamically changes parameters of machine-learning processes based on network conditions and alarm patterns. By adjusting parameters such as detection thresholds, data model configurations, and decision-making rules, the system can adapt sensitivity levels to reduce false alarms while maintaining timely detection of genuine issues.
2Measurement precision
If multiple machine-learning processes with diverse data models are used to improve recommendation accuracy, then system intelligence is improved, but system complexity increases
Solution Approach 1:
The complex system is segmented into multiple independent machine-learning processes, each with its own data model and decision-making rules. This segmentation allows each process to be relatively simple and manageable, while the collective system achieves high accuracy through diversity and aggregation of individual recommendations.
Solution Approach 2:
Multiple machine-learning processes are merged into a unified recommendation system where individual recommendations are aggregated. The merging process combines outputs from diverse data models through voting or clustering mechanisms, achieving high recommendation accuracy while managing complexity through structured integration.
3Manufacturing precision
If manual tuning and fine-tuning of configuration parameters is performed to optimize network performance, then performance precision is improved, but operation time and effort increase
Solution Approach 1:
The system performs self-service through autonomous machine-learning processes that automatically tune and optimize configuration parameters without manual intervention. The machine-learning processes learn from network data and autonomously adjust parameters to optimize performance, eliminating time-consuming manual tuning while maintaining high precision.
Solution Approach 2:
The system implements feedback mechanisms where machine-learning processes continuously monitor network performance and automatically adjust configuration parameters based on observed outcomes. This closed-loop feedback enables automatic fine-tuning that achieves high performance precision without requiring manual time and effort for parameter adjustment.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method for recommending configuration changes in a communications network. The method comprises maintaining (102) a plurality of machine-learning processes, wherein an individual machine-learning process operates based on a data model and decision- making rules, and the plurality of machine-learning processes operate based on a plurality of different data models and a plurality of different decision-making rules. The method also comprises obtaining (104) values of Key Performance Indicators, KPIs, from network elements of the communications network and obtaining a goal (106) defining at least one KPI value characterising wanted operation of the communications network. The method also comprises producing (108) by the plurality of machine-learning processes, based on the received values of KPIs and using the data models and decision-making rules, a plurality of individual recommendations; and producing an output recommendation (110) based on the produced individual recommendations. An apparatus realising the above method is also disclosed.