Generative Model Storage for Mobile Networks With Reduced Data Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile network technologies face challenges in managing the explosion of data volume for historical storage without compromising data quality or resource efficiency, particularly in 5G and beyond networks utilizing AI/ML, where existing compression techniques may lead to data loss.
Innovation Solution
Implementing generative models, such as generative adversarial networks (GANs), to generate synthetic data with similar statistical properties as the source data, reducing the volume of data stored and transmitted while maintaining quality, and utilizing these models within the analytics data repository function (ADRF) to manage data storage efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If compression techniques are used to reduce data volume for storage, then resource consumption is reduced, but data quality is lost
Solution Approach 1:
The patent creates synthetic copies of data using generative models (GANs) that replicate the statistical properties and patterns of original data. Instead of storing all raw data, the system stores synthetic data generated by trained models, dramatically reducing storage requirements while preserving data quality and enabling lossless or reduced-loss reconstruction of original data characteristics.
Solution Approach 2:
The patent transforms data from its original form into a compressed representation by training generative models on the data. The models learn key parameters and patterns, then generate synthetic data that maintains the essential statistical properties. This parameter-based representation reduces data volume while preserving quality for analytics and machine learning operations.
2Loss of information
If all historical data is stored to maintain data quality, then data quality is preserved, but resource consumption increases
Solution Approach 1:
The system creates synthetic replicas of historical data using generative models. These synthetic copies preserve the statistical properties, patterns, and relationships of the original data, enabling analytics and machine learning operations to proceed with reduced data volume while maintaining data quality requirements.
Solution Approach 2:
The patent extracts the essential information and statistical properties from large volumes of historical data by training generative models. Instead of storing all raw data, only the critical patterns and generated synthetic samples are retained, significantly reducing storage resources while preserving data quality for analytical purposes.
3Productivity
If data volume increases with 5G and AI/ML adoption, then network capabilities are enhanced, but storage and transmission resources are overwhelmed
Solution Approach 1:
The patent uses generative models to create synthetic data copies that maintain the statistical properties and patterns of original network data. This enables enhanced network capabilities for analytics and machine learning while reducing the actual data volume stored and transmitted, preventing resource overload.
Solution Approach 2:
The system transforms raw network data into compressed parameter representations through generative model training. The models capture essential network patterns and generate synthetic data that preserves analytical value while dramatically reducing data volume, enabling scalable network operations with enhanced capabilities.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There are provided measures for optimized data storage in mobile network scenarios. Such measures exemplarily comprise transmitting a data storage request requesting storage of data, said data storage request including conveyed information and an indicator indicative of a degree of involvement of generative models in said conveyed information which represents said data, and receiving a data storage acknowledge response.