Automated Intelligence Orchestration for Open RAN ML Model Deployment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Orchestrating network intelligence in Open RAN systems poses challenges due to the need for real-time adaptation of RAN parameters and the difficulty in making large amounts of input data available within tight temporal windows, as well as selecting the appropriate ML/AI models that meet performance metrics and resource requirements.
Innovation Solution
The method involves an automated intelligence orchestration framework that receives requests specifying functionality, location, and timescale, selects pre-trained ML/AI models, assigns resources, generates executable software components, and instantiates models across Open RAN resources such as non-real-time and near-real-time RICs, centralized units, and radio units, using a service management and orchestration framework.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If ML/AI models are deployed at RICs to enable real-time network control, then network intelligence and automation are improved, but the ability to make large amounts of input data available within tight temporal windows deteriorates
Solution Approach 1:
The system segments the network control architecture into multiple RIC instances (non-real-time and near-real-time RICs) with different functional capabilities. Each RIC is responsible for specific control tasks and data processing, allowing parallel operation and reducing the time burden on any single node. This segmentation enables the system to handle large data volumes by distributing processing across multiple specialized components.
Solution Approach 2:
The patent introduces an intermediary data distribution mechanism that bridges the gap between data sources and ML/AI models at RICs. This intermediary layer pre-processes, buffers, and routes data to the appropriate RIC instances, ensuring that data is available when needed without overwhelming the system. The intermediary acts as a buffer that decouples data generation from data consumption, resolving the temporal mismatch.
2Adaptability or versatility
If multiple ML/AI models are instantiated across different Open RAN resources, then network performance and adaptability are improved, but device complexity and orchestration difficulty increase
Solution Approach 1:
The system implements dynamic model instantiation and deployment across Open RAN resources based on real-time network conditions, traffic patterns, and resource availability. Models are not statically assigned but dynamically activated, migrated, or deactivated as needed. This dynamic approach allows the system to adapt to changing requirements while the orchestration system maintains awareness of the current state, managing complexity through automation rather than rigid structure.
Solution Approach 2:
The patent changes the parameter of model deployment from fixed to variable, allowing models to be instantiated at different locations (RICs, CUs, DUs) based on operational parameters such as latency requirements, data availability, and computational resources. This parameter-driven deployment strategy enables versatile network optimization while the orchestration system manages complexity by using parameter-based decision rules rather than complex hard-coded logic.
3Reliability
If ML/AI models are trained and deployed to meet specific performance metrics, then model accuracy and reliability are improved, but the difficulty of selecting appropriate models and ensuring resource adequacy increases
Solution Approach 1:
The system performs preliminary actions by pre-training and validating multiple ML/AI models before deployment, creating a repository of pre-validated models with known performance characteristics. This preliminary preparation includes training models on historical data, evaluating them against performance metrics, and storing metadata about their capabilities and resource requirements. When deployment is needed, the system selects from this pre-validated set rather than training from scratch, ensuring reliability while reducing the complexity of real-time model selection.
Data Source
AI summary
Provided herein are methods and systems for deployment and orchestration of network intelligence in an Open RAN including receiving requests at a request collector, selecting one, by an orchestration engine, or more ML/Al models applicable for satisfying the plurality of collected requests, assigning at least one Open RAN resource to execute each of the ML/Al models, automatically generating, by an orchestration engine, executable software components embedding at least one of the ML/Al models, dispatching each executable software component to the assigned Open RAN resource, and instantiating, at the Open RAN resource, at least one of the ML/AI models embedded in the executable software component to configure the Open RAN to satisfy the requests.