Inference Data Selection for Low-Load RAN Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently performing learning due to the need for data processing and network load when collecting learning data from external servers and combining it with RAN data, which can overwhelm the Non-RT RIC and network resources.
Innovation Solution
A system and method that involves acquiring inference data from a data providing apparatus, specifying data collected from another system for learning, and transmitting selected data as learning data to a learning model, thereby reducing data processing and network load by optimizing data collection and transfer schemes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If learning data is collected from external servers and combined with RAN data, then learning can be performed, but the load on Non-RT RIC and network resources increases
Solution Approach 1:
The patent segments the learning data collection process by identifying and separating inference data from different systems (RAN and external servers). The Non-RT RIC selectively collects only necessary inference data from external servers and combines it with RAN data, dividing the data collection task into manageable parts that reduce overall system load while maintaining learning effectiveness.
Solution Approach 2:
The patent extracts only the essential inference data needed for learning from external servers, rather than collecting all available data. The Non-RT RIC identifies and extracts relevant inference data that will contribute to learning model construction, eliminating unnecessary data transfer and processing overhead on network resources.
2Reliability
If all inference data is collected and processed, then comprehensive learning can be achieved, but network load and data processing time increase
Solution Approach 1:
The patent applies preliminary action by having the Non-RT RIC identify and select necessary inference data before the learning process begins. The system pre-processes and filters inference data from external servers to determine which data will be useful for learning model construction, avoiding the time-consuming process of collecting and then filtering all available data later.
3Measurement precision
If inference data from multiple sources is combined, then learning model accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent applies local quality by treating inference data from different sources (RAN and external servers) differently based on their specific characteristics and relevance to the learning task. The Non-RT RIC selectively processes and combines data from each source according to its quality and suitability for specific learning objectives, rather than applying uniform processing to all data.
Data Source
AI summary
A first system (10) includes an acquisition unit (11) that acquires data provided from a data providing apparatus such as an external server as inference data for a second system (20) to perform inference by an inference model, and a specifying unit (12) that specifies, from among data including the inference data 5acquired by the acquisition unit (11), data collected from the second system (20) that has performed inference by the inference model, as learning data for a learning model for constructing the inference model.