Radio Access Network Data Model Sharing with Compact Representation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data sharing methods in radio access networks face challenges such as high reporting overhead, privacy concerns, and the need for mutual agreements on AI/ML techniques, which can hinder efficient data model sharing and prediction performance.
Innovation Solution
A method where network nodes request and share data models with compact representations and correlation parameters, allowing for accurate data modeling without transmitting large volumes of raw data, enabling flexibility across different network vendors and protecting user and network privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If raw data is transmitted between network nodes for machine learning analysis, then prediction accuracy is improved, but reporting overhead increases and privacy concerns arise
Solution Approach 1:
The patent extracts only the essential information needed for machine learning analysis by generating compact representations (summaries, statistics, or processed data) instead of transmitting raw data. This allows network nodes to share meaningful patterns and insights while eliminating unnecessary data volume, thereby reducing reporting overhead and protecting privacy while maintaining prediction accuracy.
Solution Approach 2:
The patent creates a simplified copy or representation of the raw data that captures the essential characteristics needed for analysis. This compact representation (such as aggregated statistics or processed features) serves as a substitute for the full raw data, enabling machine learning operations without transferring the complete data set, thus reducing data quantity while preserving analytical value.
2Adaptability or versatility
If network nodes share data models for machine learning, then collaboration is improved, but complexity of coordinating AI/ML techniques increases
Solution Approach 1:
The patent standardizes data model sharing by transforming and normalizing parameters into a common framework. This allows different network nodes with varying AI/ML techniques to collaborate by converting their models into a unified parameter representation, thereby enabling versatility and collaboration while reducing the complexity of coordinating diverse techniques through standardization.
3Measurement precision
If data models are calculated and shared between network nodes, then prediction performance is improved, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary calculations and pre-processing of data locally at each network node before sharing. By preparing compact representations and essential features in advance, nodes reduce the computational burden during data sharing and model calculation phases, thereby improving prediction performance while minimizing processing time and resource consumption.
Data Source
AI summary
Methods performed by network nodes in a radio access network may be provided. A first network node may receive an identification of data models supported by a second network node. Each of the identified data models may include a data model for calculating a compact representation of data collected, at least one correlation parameter, and an accuracy metric. The first network node may transmit a request that the second network node calculate an identified one of the data models. The first network node may receive the calculated data model from the second network node. The first network node may evaluate the received calculated data model based on determining whether the received calculated data model correlates to at least one correlation parameter.