ML Model Training for Cell Performance Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The effective and efficient design, operation, and troubleshooting of 5G and next-generation mobile networks are hindered by the challenge of training accurate machine-learning (ML) models, which require large volumes of labeled training data and are costly to develop.
Innovation Solution
An automated system that uses network measurements and key performance indicators (KPIs) as labeled training data to build accurate ML models for predicting cell performance in communication networks. This system iteratively executes a training procedure for the ML model based on comparing predicted and current cell performance, resulting in a locally trained model that provides low-latency responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If large volumes of labeled training data are used to train ML models, then prediction accuracy is improved, but training cost and complexity increase
Solution Approach 1:
The system performs preliminary actions by automatically generating labeled training data from network measurements and KPIs before the actual model training process. This pre-preparation of high-quality labeled data reduces the overall training complexity and cost while maintaining prediction accuracy, as the data preparation work is done systematically in advance rather than manually during the training process
Solution Approach 2:
The system enables self-service by using automated processes to generate labeled training data from existing network operations data. The ML model trains itself using the automatically generated labels without requiring extensive manual annotation, reducing human intervention and training complexity while maintaining data quality and prediction accuracy
2Loss of time
If ML models are trained locally at network edges, then response latency is reduced, but data preparation and model training resources are required at distributed locations
Solution Approach 1:
The system segments the ML model training process by distributing trained models to multiple network edge locations where they can operate independently. This segmentation allows local predictions at the edge without requiring centralized processing, reducing response latency while the initial centralized training phase handles the complexity of data preparation and model development
Solution Approach 2:
The system creates copies of the trained ML model and deploys them to multiple network edge locations. Instead of requiring complex distributed training infrastructure at each edge location, the model is trained centrally and then copied to edges for local execution, achieving low latency while avoiding the complexity of distributed training systems
Data Source
AI summary
Aspects of the subject disclosure may include, for example, a method performed by a processing system; the method includes receiving a plurality of values of key performance indicators (KPIs) relating to performance of a cell on a communication network. The plurality of values of the KPIs includes labeled training data for training a machine learning (ML) model for the performance of the cell. The method further includes iteratively executing, using the labeled training data, a training procedure for the ML model; and testing the trained ML model. The labeled training data corresponds to ground truth data that may include a training data set, a validation data set and a test data set. The trained ML model, when deployed on a communication network, receives as input near-real time data regarding the performance of the cell and provides as output predictions of the performance of the cell. Other embodiments are disclosed.