Adaptive AI Model Selection for Wireless Channel State Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems, particularly in 5G and 4G LTE environments, face challenges in dynamically adapting AI models within wireless transmit/receive units (WTRUs) to optimize processing based on changing computational resources and environmental conditions, leading to suboptimal performance in terms of power consumption, latency, and inference accuracy.
Innovation Solution
The WTRU adapts its AI model by determining triggering conditions related to processing capability changes, HARQ NACKs, bandwidth part changes, and other environmental factors to switch between different data processing models, such as AI, ML, or DL models, optimizing parameters like model structure, layer configuration, and quantization levels to improve performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed AI model is used in WTRU, then implementation is simple, but performance is suboptimal under varying computational resources and environmental conditions
Solution Approach 1:
The patent implements dynamic AI model adaptation by enabling the WTRU to switch between different data processing models (AI, ML, DL) and adjust model parameters based on real-time triggering conditions such as processing capability changes, HARQ NACK rates, and bandwidth part changes. This transforms the static model into a dynamic system that adapts to varying computational resources and environmental conditions, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The patent changes model parameters including model structure, layer configuration, and quantization levels based on triggering conditions. By dynamically adjusting these parameters, the system achieves adaptability to different operational scenarios while managing complexity through structured parameter modification rather than complete model redesign.
2Measurement precision
If a complex AI model is used to improve inference accuracy, then processing power requirements increase, but available computational resources are limited
Solution Approach 1:
The patent dynamically adjusts model parameters including quantization levels and layer configuration based on available processing capability. When resources are abundant, higher precision models with finer quantization are used to maximize inference accuracy. When resources are constrained, the system reduces model complexity and quantization precision, thereby managing power consumption while adapting to available computational resources.
Solution Approach 2:
The system dynamically selects between different data processing models (AI, ML, DL) and adjusts model characteristics based on real-time processing capability assessments. This dynamic adaptation allows the system to optimize the balance between inference accuracy and power consumption by matching model complexity to available resources at any given moment.
3Adaptability or versatility
If AI model parameters are frequently adjusted to optimize performance, then adaptability improves, but processing latency increases
Solution Approach 1:
The patent implements periodic model adaptation triggered by specific events such as bandwidth part changes, processing capability changes, or HARQ NACK rate thresholds. Rather than continuous adjustment, the system periodically evaluates triggering conditions and adapts models only when necessary, reducing unnecessary processing overhead and latency while maintaining adaptability to significant environmental changes.
Solution Approach 2:
The system uses feedback mechanisms by monitoring HARQ NACK rates and processing capability changes to trigger model adaptations. This feedback-driven approach ensures adaptations occur only when performance degradation is detected or conditions warrant change, optimizing the balance between adaptability and latency by avoiding premature or unnecessary model adjustments.
Data Source
AI summary
A wireless transmit/receive unit (WTRU) may adapt an AI model, e.g., based on changes in computational resources, changes in a power status, etc. For example, a WTRU may determine first channel state information (CSI) feedback information using a first data processing model. The WTRU may determine that a triggering condition associated with use of the first data processing model has been met. The WTRU may determine, based on the determination that the triggering condition has been met, a data processing model to use to determine second CSI feedback information, where the data processing model is different than the first data processing model. The WTRU may transmit an indication of the determined data processing model. The WTRU may determine the second CSI feedback information using the determined data processing model. The WTRU may transmit an indication of the determined second CSI feedback information.