AI-Driven Radio Reset Selection to Reduce Dropped Calls
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing radio reset strategies in Radio Access Networks (RANs) often lead to unnecessary operational disruptions, such as dropped calls, due to the difficulty in identifying the specific radio causing issues, leading to inefficient and reactive recalibration measures.
Innovation Solution
Implementing an AI-driven radio reset selection system using machine learning (ML) models to analyze radio data, convert it into vector embeddings, and compare them to a vector database to identify the most likely problematic radios, recommending or automatically performing soft or hard resets based on AI/ML models trained for radio problem isolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a group of radios is rested to force recalibration, then the operational issue is mitigated, but dropped calls and service disruptions occur
Solution Approach 1:
The system segments the radio population by creating individual vector embeddings for each radio based on their unique operational characteristics and performance data. This allows the AI model to identify and target specific problematic radios rather than applying blanket resets to entire groups, thereby isolating the harmful effect to only the necessary radios and minimizing dropped calls.
Solution Approach 2:
The system performs preliminary analysis by continuously collecting radio performance data, creating vector embeddings, and training AI models to predict which radios are likely to fail or cause issues before they do. This proactive identification allows operators to perform targeted resets only when and where needed, preventing dropped calls that would occur from unnecessary preventive resets of functioning radios.
2Ease of operation
If manual radio reset selection is performed, then operational control is maintained, but time consumption and inefficiency increase
Solution Approach 1:
The system enables self-service by automatically collecting radio performance data, generating vector embeddings, training AI models, and identifying problematic radios without requiring manual expert analysis. The AI-driven platform autonomously performs the complex task of radio problem isolation, significantly reducing the time and expertise needed for operators to identify which radios require resetting.
Solution Approach 2:
The system replaces manual mechanical analysis methods with AI-driven automated analysis. Instead of operators manually examining radio performance data and making decisions about which radios to reset, the system uses machine learning models to automatically analyze vector embeddings and identify problematic radios, dramatically reducing operational time and improving efficiency.
3Measurement precision
If AI/ML models are trained for radio problem isolation, then accurate radio identification is achieved, but system complexity increases
Solution Approach 1:
The system introduces vector embeddings as an intermediary representation layer between raw radio performance data and AI model analysis. By converting complex multi-dimensional radio data into standardized vector embeddings, the system simplifies the input for AI models while maintaining high identification accuracy. This intermediary step makes the overall system more manageable despite the complexity of the AI/ML components.
Data Source
AI summary
Artificial intelligence (AI)-driven radio reset selection to mitigate operational impacts is disclosed. Such an approach may minimize or avoid potentially unnecessary operational rests on the radios. This enhances the end user experience by minimizing or avoiding dropped calls and performs preventive measures instead of reactive measures. Radios are rested or cell sites are brought back to steady states before a major failure occurs.