Sub-Flow Differentiation for AI/ML Data Priority Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In wireless communication networks, the exchange of AI/ML model parameters between user equipment (UE) and base stations involves a high volume of data, including redundant data, which can lead to inefficient processing and increased air interface overhead.
Innovation Solution
The method involves differentiating important data sub-flows from less important data sub-flows by assigning them different priorities and transmission configurations, using a hyper-frame format to ensure higher reliability for important data, thereby reducing air interface overhead while maintaining AI/ML model performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If all data is transmitted with equal priority, then transmission simplicity is maintained, but air interface overhead increases and processing efficiency decreases
Solution Approach 1:
The patent segments the data flow into multiple sub-flows based on importance levels (e.g., critical data vs. non-critical data). This segmentation allows different transmission strategies to be applied to different data portions, improving processing efficiency by handling critical data with higher priority while reducing overhead for less important data.
Solution Approach 2:
The patent applies local quality by assigning different transmission priorities and configurations to different data sub-flows. Critical data receives higher priority transmission with better error protection, while non-critical data uses best-effort transmission. This localized differentiation optimizes resource allocation without requiring system-wide complexity.
2Reliability
If redundant data is transmitted to ensure AI/ML model performance, then model training accuracy is maintained, but air interface overhead increases
Solution Approach 1:
The patent extracts and identifies critical data elements from the overall data flow that are essential for AI/ML model training. By separating these critical elements from redundant data, the system can transmit only the necessary information with high reliability while reducing overall air interface overhead through selective transmission of non-critical data.
Solution Approach 2:
The patent changes transmission parameters (priority level, error protection, retransmission settings) based on data importance. Critical data uses parameters optimized for reliability (lower priority thresholds, better error correction), while non-critical data uses parameters optimized for efficiency. This parameter differentiation maintains model performance while reducing energy consumption.
3Reliability
If higher priority is given to all data, then transmission reliability is improved, but transmission speed and spectral efficiency decrease
Solution Approach 1:
The patent segments data into priority levels and applies differentiated transmission strategies. Critical data receives high reliability transmission with appropriate error protection and retransmission mechanisms, while non-critical data is transmitted at best effort speed. This segmentation allows the system to maintain reliability for essential data without sacrificing overall transmission speed.
Solution Approach 2:
The patent applies partial action by providing enhanced reliability only to the extent necessary for critical data, rather than uniformly to all data. This selective application of reliability mechanisms maintains transmission speed for non-critical data while ensuring reliability where it matters most.
Data Source
AI summary
A method includes receiving a packet including: a first part, where the first part of the packet is represented by bits of a first class, and a second part, where the second part of the packet is represented by bits of a second class. The method further includes transmitting the bits of the first class with a first air interface configuration, and transmitting the bits of the second class with a second air interface configuration.