UE ML Capability Reporting for 5G Adaptability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems, particularly 5G NR, lack efficient mechanisms for reporting and managing machine learning (ML) capabilities between user equipment (UE) and base stations, which can impact the performance and adaptability of ML procedures in these systems.
Innovation Solution
The proposed solution involves a method where user equipment (UE) and base stations can query and report ML capabilities, perform ML procedures based on initial and updated capabilities, and manage these capabilities through scheduling requests and grants, ensuring seamless operation and adaptation of ML processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If ML capabilities are reported and updated dynamically between UE and base station, then adaptability and performance of ML procedures are improved, but device complexity and signaling overhead increase
Solution Approach 1:
The patent implements dynamic ML capability reporting where the UE can update its ML capabilities (such as supported ML models, computational resources, memory availability) at different times rather than reporting static capabilities only during initial connection. This allows the network to adapt to changing UE conditions and improve ML procedure performance while maintaining a structured update mechanism.
Solution Approach 2:
The patent establishes a feedback loop where the base station queries UE ML capabilities, the UE reports its capabilities, and the network uses this information to configure appropriate ML procedures. This feedback mechanism enables continuous adaptation of ML operations based on actual UE state, resolving the contradiction between adaptability and complexity through systematic information exchange.
2Loss of information
If ML capability updates are transmitted frequently to maintain current capability information, then information accuracy is improved, but loss of time and signaling overhead increase
Solution Approach 1:
The patent implements periodic capability reporting where the UE updates its ML capabilities at scheduled intervals or when triggered by specific events (such as changes in computational resources or supported models). This periodic mechanism ensures the network has relatively current capability information without requiring continuous real-time reporting, thus balancing information accuracy with time efficiency.
Solution Approach 2:
The patent allows the UE to proactively report capability updates before the network requests them, particularly when significant changes occur in ML-related resources. This preliminary action ensures the network has accurate capability information when needed without wasting time on unnecessary update cycles, resolving the contradiction between information freshness and time consumption.
Data Source
AI summary
The present disclosure relates to methods and devices for wireless communication of an apparatus, e.g., a UE and/or a base station. The apparatus may perform a ML procedure based on at least one initial ML capability of the UE. The apparatus may also determine at least one updated ML capability for the ML procedure corresponding to an update to the at least one initial ML capability. Further, the apparatus may transmit, to a base station, an indication of the at least one updated ML capability for the ML procedure. The apparatus may also perform the ML procedure based on the at least one updated ML capability or based on the at least one initial ML capability.