Cross-Node AI/ML Sessions in Cloud RAN for Latency Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems face challenges in efficiently managing AI/ML processing in radio access networks (RAN), leading to computational resource constraints and increased costs for RAN operators.
Innovation Solution
The implementation of cross-node AI/ML sessions between user equipment (UE) and RAN controllers in a cloud-based RAN architecture, which configures entities for efficient AI/ML processing, reduces processing latencies, and distributes resources across a cloud platform.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI/ML processing is implemented in RAN, then wireless communication performance is improved, but computational resource constraints and costs increase
Solution Approach 1:
The patent segments AI/ML processing into cross-node sessions between multiple network entities (UE, gNB, O-RAN controller). This distributes computational load across segmented nodes rather than concentrating it in one entity, thereby improving overall system performance while managing resource consumption through distributed processing.
Solution Approach 2:
The O-RAN controller acts as an intermediary that manages and coordinates AI/ML processing between UEs and gNBs. It receives machine learning information from gNBs, processes it, and generates output data that is transmitted back, thereby mediating the computational workload and reducing direct processing demands on individual nodes.
2Loss of time
If cross-node AI/ML sessions are implemented, then processing latency is reduced, but network complexity increases
Solution Approach 1:
The system establishes cross-node AI/ML sessions in advance between network entities and UEs. Configuration messages are exchanged beforehand to set up the processing framework, so that when actual AI/ML processing is needed, the infrastructure is already in place, reducing processing latency while managing complexity through pre-planned configurations.
Solution Approach 2:
The O-RAN controller provides universal functionality by managing multiple AI/ML sessions simultaneously between different UEs and gNBs. It handles diverse machine learning information types and coordinates various processing tasks, thereby reducing per-session overhead and latency while amortizing the complexity across multiple universal operations.
3Quantity of substance
If machine learning information is exchanged between network entities, then network capacity is enhanced, but signaling overhead increases
Solution Approach 1:
The patent extracts and separates machine learning information from regular communication data streams. Dedicated signaling messages are used to transport machine learning information, configuration data, and output results between entities. This extraction allows efficient handling of AI/ML data without interfering with primary communication functions, enhancing network capacity while managing signaling overhead through specialized information channels.
Data Source
AI summary
Certain aspects of the present disclosure provide techniques for performing cross-node machine learning operations in a radio access network. An example method of wireless communication by a first network entity includes providing, to a second network entity, an indication of cross-node machine learning information used for a cross-node machine learning session between the first network entity and a user equipment (UE); obtaining machine learning information associated with the UE; and controlling the cross-node machine learning session based at least in part on the machine learning information.