RAN Ensemble ML Model Provisioning for Device-Specific Wireless Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for providing Machine Learning (ML) models to wireless devices in communication networks incur significant resource costs and inefficiencies due to unicast transmissions of individually tailored models, which are not adaptable to dynamic device requirements and capabilities.
Innovation Solution
A method involving a Radio Access Network (RAN) node that transmits a plurality of base ML models trained using ensemble methods, allowing wireless devices to select and combine models based on their specific needs and capabilities, reducing resource overhead through broadcast or multicast transmissions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If unicast transmission of individually tailored ML models is performed to each device, then model accuracy and device-specific optimization are improved, but network resource consumption and transmission overhead increase significantly
Solution Approach 1:
The patent segments a large ML model into multiple smaller base models that can be independently transmitted and combined. This allows the network to send only necessary model components to devices, reducing transmission overhead while maintaining model accuracy through selective combination of base models at the device side.
Solution Approach 2:
The patent creates a universal framework where a set of base models can serve multiple devices with different requirements. By transmitting a common set of base models that can be selectively combined, the system achieves device-specific optimization without requiring separate unicast transmissions for each device, thus reducing network resource consumption.
2Adaptability or versatility
If unicast transmission of individually tailored ML models is performed to each device, then device-specific QoS requirements are met, but transmission time and network overhead increase
Solution Approach 1:
The patent performs preliminary segmentation of the ML model into base models and prepares combination configurations in advance. This allows devices to quickly select and combine pre-prepared base models according to their QoS requirements, avoiding the need for time-consuming custom model generation and transmission for each device.
Solution Approach 2:
The patent uses a copying mechanism where a universal set of base models is transmitted once and then copied/combined in different configurations at the device side to meet specific QoS requirements. This eliminates the need for multiple original model transmissions while still providing device-specific adaptations.
3Productivity
If ML models are transmitted to devices, then device processing capabilities are utilized, but device energy consumption and processing overhead increase
Solution Approach 1:
The patent enables dynamic selection and combination of base models at the device side based on current processing capabilities and energy status. Devices can adaptively choose which base models to activate and combine, allowing flexible adjustment between processing utilization and energy consumption according to real-time conditions.
Solution Approach 2:
The patent allows devices to change model parameters by selectively combining different base models based on their processing capabilities and energy constraints. This enables the same set of base models to be configured with different effective parameters for different devices, optimizing the balance between processing utilization and energy consumption.
Data Source
AI summary
A method for managing a plurality of wireless devices. The method includes obtaining a plurality of base Machine Learning (ML) models, wherein each base ML model is operable to provide an output on the basis of which at least one RAN operation performed by a wireless device may be configured. The method further includes transmitting characterising information for individual models of the plurality of base ML models and configuration information for the plurality of base ML models over the RAN. The method further includes receiving an indication of which one or more of the plurality of base ML models the wireless device will be using as an ensemble ML model in connection with a RAN operation performed by the wireless device, and setting a value of at least one configuration parameter associated with the RAN operation performed by the wireless device based on the received indication.