Disaggregated RAN Control via ML Network Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication and data networks lack visibility and optimization at a macro or network-wide perspective, with control plane elements making suboptimal decisions due to limited visibility and interconnection capabilities.
Innovation Solution
A method and apparatus for managing and controlling communication networks by obtaining models, applying data to generate updated controls, and processing parametric data to optimize communications, enabling real-time adjustments and pattern identification for improved network performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If control plane elements make decisions based on localized cell-level information, then local communication control is effective, but network-wide optimization and visibility are insufficient
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between localized control plane elements and network-wide optimization. The model receives parametric data from multiple control plane elements and generates optimized control decisions that incorporate network-wide patterns, effectively mediating between local control needs and global optimization requirements.
Solution Approach 2:
The system implements a feedback loop where parametric data from control plane elements is continuously collected, processed through machine learning models, and used to generate updated control decisions. This feedback mechanism enables network-wide optimization by learning from aggregated local decisions and applying improvements across the network.
2Adaptability or versatility
If control decisions are made independently at each control plane element, then local autonomy is maintained, but suboptimal decisions occur due to lack of macro perspective
Solution Approach 1:
The patent segments the control function into two parts: local control plane elements that maintain autonomy and a machine learning model that provides network-wide optimization guidance. This segmentation allows local elements to maintain their adaptive decision-making while incorporating improved decisions from the centralized model, resolving the contradiction between autonomy and decision quality.
Solution Approach 2:
The system performs preliminary analysis of parametric data from multiple control plane elements to identify patterns and generate optimized control strategies before applying them. This preliminary action allows the machine learning model to prepare improved decisions that can be applied across the network, enhancing decision quality while preserving local autonomy.
3Device complexity
If cell-by-cell interconnections are used for neighborhood relations, then local network control is achieved, but macro or network-wide perspective is not directly addressed
Solution Approach 1:
The patent creates a virtual copy of the network state through machine learning models that process parametric data from control plane elements. Instead of requiring direct macro-level interconnections, the system creates computational copies of network behavior and uses these to generate optimization strategies, thereby achieving network-wide visibility without increasing physical interconnection complexity.
Data Source
AI summary
Aspects of the subject disclosure may include, for example, obtaining a model of a communication network, applying first data to the model to generate an updated model, processing the updated model to establish a first control, enabling first communications in the communication network in accordance with the first control, obtaining first parametric data regarding the first communications in the communication network that are based on the first control, processing the first parametric data in accordance with the updated model to generate an updated first control, and enabling second communications in the communication network in accordance with the updated first control. Other embodiments are disclosed.