RAN CNF Portability via Dynamic Task Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Radio Access Network (RAN) systems face vendor lock-in issues due to tightly coupled DU software with specific hardware platforms, limiting portability and efficient resource utilization across different hardware architectures.
Innovation Solution
A system and method for developing and deploying Radio Access Network Containerized Network Functions (RAN CNF) using a Software Development Kit (SDK) that enables platform-independent expression of RAN functionality as a pipeline of tasks, along with a schedule generator and scheduler runtime unit, allowing allocation and loading of tasks across multiple hardware platforms without source code changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If DU software is tightly coupled with specific hardware platforms using proprietary SDK APIs, then the software can be optimized for that specific platform, but the software becomes non-portable to other hardware platforms resulting in vendor lock-in
Solution Approach 1:
The patent segments the DU software into independent functional modules called RAN tasks that can be separately developed, scheduled, and executed on different hardware platforms. This segmentation allows the software to be ported across platforms by reconfiguring task allocation rather than rewriting platform-specific code.
Solution Approach 2:
The patent introduces a platform-independent task scheduling layer as an intermediary between the RAN software and underlying hardware platforms. This scheduling layer abstracts hardware-specific details, enabling the same RAN task code to run on different hardware architectures without modification.
2Reliability
If computing cores are assigned to L1 RAN tasks statically, then the system ensures predictable performance, but the computing resources are not optimally utilized
Solution Approach 1:
The patent transitions from static to dynamic task scheduling where the scheduling layer can dynamically allocate and reassign RAN tasks to different computing cores based on current system conditions, workload characteristics, and performance requirements. This dynamic approach maintains performance predictability through scheduling policies while optimizing resource utilization.
Solution Approach 2:
The patent changes the scheduling parameter from fixed static assignment to dynamic adjustable allocation. The system can modify task-to-core mappings in real-time based on performance metrics and resource availability, enabling both predictable performance through controlled parameter changes and improved utilization through adaptive optimization.
3Manufacturing precision
If multiple software versions are maintained for multiple hardware platforms, then each platform receives optimized support, but the development and maintenance complexity increases significantly
Solution Approach 1:
The patent creates a universal RAN task software version that can execute on multiple hardware platforms through the platform-independent scheduling layer. This single universal version replaces multiple platform-specific versions, eliminating the need to maintain separate codebases while still providing optimized performance through the scheduling system's ability to adapt task allocation to platform characteristics.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A system for deploying a Radio Access Network Containerized Network Function (RAN CNF) that is portable across a plurality of RAN hardware platforms is provided. The system includes a Software Development Kit (SDK) (206), a schedule generator (214) and a scheduler runtime unit (216). The SDK enables providing a RAN functionality in a physical layer (L1) software code (306) in a platform-independent manner as a RAN pipeline of a plurality of RAN tasks. The RAN tasks include a first and second RAN task. The first RAN task invokes an Application programming interface (API) from a plurality of Application Programming Interfaces (208A-N) to call to the second RAN task. The schedule generator (214) generates a schedule for allocating a node in the RAN pipeline to one or more processing elements (218A-N). The scheduler runtime unit (216) loads the RAN tasks corresponding to nodes in the RAN pipeline, based on the schedule generated by the schedule generator (214).