Communication Network Parameter Changes Using De-Confused Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network operators face challenges in accurately configuring carrier parameters for optimal performance due to the large number of network parameters and complex dependencies across different locations, user behaviors, and varying signal propagation patterns, making it difficult to systematically document and implement common network parameter changes across a communication network.
Innovation Solution
A data-driven machine learning approach using a classifier trained on a dataset of network parameter changes, applying a de-confusion process to reduce impact confusion and automatically implement network parameter changes based on network-wide records, leveraging a SON orchestrator to optimize network performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If network operators manually configure carrier parameters for each location, then network performance can be optimized for specific conditions, but the complexity and time required for configuration increases significantly
Solution Approach 1:
The system creates copies of successful network parameter change records from one location and applies them to similar locations. The classifier learns from historical records of parameter changes that improved network performance, automatically copying these configurations to new carriers based on similarity in attributes such as signal propagation patterns and traffic characteristics.
Solution Approach 2:
The system automatically changes network parameters by selecting from pre-defined parameter change groups stored in the database. Instead of manual configuration, the classifier determines which parameter changes to apply based on learned patterns from historical data, automatically adjusting parameters like handover thresholds, load balancing settings, and interference management configurations.
2Reliability
If network operators document and implement parameter changes manually, then performance improvements can be captured, but the process is time-consuming and error-prone
Solution Approach 1:
The system performs self-service by automatically learning from historical network parameter changes and their performance impacts. The classifier autonomously identifies which parameter changes to apply to new carriers without human intervention, automatically querying the database for suitable parameter change groups and applying them based on learned decision rules.
Solution Approach 2:
The system performs preliminary action by pre-processing and storing network parameter change records in a structured database before they are needed. Historical parameter changes are documented, categorized into parameter change groups, and stored with their performance outcomes, enabling rapid retrieval and application when new carriers are deployed.
3Ease of operation
If the system applies the same parameter changes across all network locations, then implementation is simplified, but network performance optimization is reduced due to varying local conditions
Solution Approach 1:
The system applies local quality by tailoring parameter changes to specific local conditions at each network location. The classifier evaluates attributes specific to each carrier and location, such as signal propagation patterns, traffic characteristics, and interference levels, to determine the most appropriate parameter changes from the database, ensuring local optimization rather than uniform application.
Solution Approach 2:
The system segments the network into distinct locations with unique characteristics, treating each carrier and location as a separate entity requiring customized parameter configuration. Rather than applying a single universal configuration, the system divides the parameter space into location-specific segments based on learned similarities and differences in network conditions.
Data Source
AI summary
A processing system may obtain a data set with records of network parameter changes, each record including at least one network parameter change and at least one attribute associated with a first aspect of a communication network, and a corresponding network performance indicator change. A first record may include a plurality of network parameter change groups. The processing system may next perform a de-confusion process by identifying a second record comprising a single network parameter change group, determining that a corresponding network performance indicator change is different from that of the at least the first record, and updating the data set to replace the first record with at least two replacement records. The processing system may apply at least one of the network parameter change groups to a second aspect of the communication network based upon a decision output of a classifier that is trained using the updated data set.