Network Function Codelet Generation With LLM Constraint Verification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Generating custom codelets for network functions in virtualized radio access networks is challenging due to the need for specialized knowledge and the difficulty in ensuring compliance with execution constraints, particularly in open-RAN environments with multiple vendors.

Innovation Solution

Utilizing a large language model (LLM) to generate codelets through a codelet generation application that receives natural language intents, generates prompts, and verifies the procedural language code to ensure compliance with network function constraints.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If dedicated processing hardware is deployed for each base station, then processing performance and reliability are improved, but deployment cost and device complexity increase

Engineering Contradiction:
Improvebase station processing reliabilityVSAvoiddedicated processing hardware complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent creates a virtual copy of the base station processing function through a codelet that can be instantiated on generic computing resources. Instead of deploying physical dedicated hardware for each base station, the system copies the processing functionality as software that can run on shared infrastructure, eliminating the need for separate dedicated hardware while maintaining processing capabilities.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical/dedicated hardware system with a software-based virtualized system. The codelet is a software module that can be deployed on generic computing platforms, substituting physical dedicated hardware with virtual processing resources that can be dynamically allocated and managed through software rather than requiring complex physical infrastructure for each base station.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Device complexity

If virtualized RAN with generic computing resources is used, then deployment cost and device complexity are reduced, but codelet generation difficulty and verification complexity increase

Engineering Contradiction:
Improveprocessing hardware complexityVSAvoidcodelet generation ease
Core Design Contradiction:
Device complexityVSEase of manufacture

Solution Approach 1:

The system provides self-service capabilities through automated codelet generation and verification. The platform automatically generates codelets from high-level specifications and performs verification without requiring manual intervention from developers, allowing users to create codelets through simple interfaces while the system handles the complex generation and validation processes automatically.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent implements automated verification that provides feedback on codelet correctness, safety, and performance before deployment. The verification system analyzes generated codelets and returns feedback information that can be used to refine and improve the codelet generation process, ensuring that generated codelets meet required specifications without manual verification by developers.

Inventive Principle:
Principle #23Feedback

3Manufacturing precision

If manual codelet generation is performed, then codelet customization and precision are improved, but development time and skill requirement increase

Engineering Contradiction:
Improvecodelet customization precisionVSAvoidcodelet development time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by automatically generating codelet frameworks, templates, and verification configurations before the actual codelet implementation. The platform pre-provides standardized structures and verification rules that guide the codelet generation process, allowing developers to focus only on the specific customization requirements rather than building everything from scratch, thereby reducing development time while maintaining precision.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent enables parameter changes by allowing customization of codelet behavior through configuration parameters and high-level specifications rather than requiring manual coding. Users can modify parameters such as data types, processing logic, and verification criteria through intuitive interfaces, and the system automatically generates the corresponding code, maintaining precision while significantly reducing development effort and time.

Inventive Principle:
Principle #35Parameter changes

4Reliability

If static verification is performed on codelets, then execution safety and reliability are improved, but verification complexity and processing time increase

Engineering Contradiction:
Improvecodelet execution safetyVSAvoidverification system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The verification system is segmented into multiple independent verification stages that can be executed sequentially. Instead of a single complex verification process, the system divides verification into separate checks for syntax correctness, semantic validity, safety constraints, and performance requirements. Each stage operates independently and can be optimized separately, reducing the overall complexity of the verification system while maintaining comprehensive safety checks.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20250362887A1Generation of codelets for network functions based on large language model
Publication Date: 2025.11.27 MICROSOFT TECHNOLOGY LICENSING LLC
  • US20250362887A1 patent drawing
  • US20250362887A1 patent drawing
  • US20250362887A1 patent drawing

AI summary

Example implementations include a method, apparatus, and computer-readable medium configured for generating codelets. An interface between a user and a large language model receives a natural language intent for a codelet to be executed within a network function to output one or more requested pieces of information from the network function. A codelet generation application generates a prompt to the large language model to write a codelet in an imperative procedural language to output the one or more requested pieces of information, the prompt including a hook point of the network function and one or more data fields of the network function associated with the one or more requested pieces of information. The application receives, in response to the prompt, imperative procedural language code for the codelet. The application statically verifies whether the procedural language code for the codelet satisfies constraints for execution within the network function.