ML Beam Selection for 5G QoS Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current beam selection methods in 5G NR wireless communication systems are inefficient when handling multiple UEs and varying QoS classes, leading to scheduling delays and suboptimal resource allocation due to reliance on traditional measurement-based approaches.
Innovation Solution
Implementing machine learning models, such as Q-learning and deep Q-networks, for data radio bearer specific beam selection, which considers throughput and latency critical factors to optimize beam choice beyond traditional signal quality metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional measurement-based beam selection is used, then implementation complexity is low, but network capacity and resource allocation efficiency are insufficient
Solution Approach 1:
The patent replaces traditional measurement-based beam selection mechanisms with machine learning-based selection. The ML model processes multiple input parameters including signal quality metrics, channel state information, and QoS requirements to dynamically determine optimal beams, thereby increasing network capacity and resource allocation efficiency while managing system complexity through intelligent automation
Solution Approach 2:
The patent introduces multiple input parameters to the beam selection process including signal quality measurements, channel state information, and QoS class requirements. By changing from single-metric measurement to multi-parameter ML-based decision making, the system achieves more efficient resource allocation and higher network capacity across different service types
2Loss of time
If traditional measurement-based beam selection is used, then implementation is simple, but scheduling delays occur and latency is increased
Solution Approach 1:
The patent performs preliminary beam evaluation and classification by pre-defining QoS classes and their associated parameters. The ML model is trained in advance to recognize patterns in channel state information and signal quality measurements, enabling rapid beam selection decisions during actual scheduling without requiring complex real-time calculations, thus reducing scheduling delays
Solution Approach 2:
By substituting traditional step-by-step measurement and selection procedures with a trained machine learning model, the system achieves faster beam selection decisions. The ML model processes multiple input parameters simultaneously and outputs optimal beam choices instantly, eliminating the sequential measurement and evaluation steps that cause scheduling delays in traditional systems
3Adaptability or versatility
If uniform beam selection is used for all UEs, then system complexity is reduced, but QoS requirements for different data types cannot be met
Solution Approach 1:
The patent applies different beam selection strategies for different QoS classes. The system identifies whether data belongs to latency-critical or best-effort categories and selects beams accordingly based on local requirements. This allows tailored beam selection for each UE's specific QoS needs while using a unified ML framework, achieving adaptability without excessive complexity
Solution Approach 2:
The patent implements dynamic beam selection that adapts to varying QoS requirements in real-time. The ML model continuously receives updated channel state information and QoS parameters, dynamically adjusting beam choices based on current network conditions and UE requirements. This dynamic adaptation enables the system to meet diverse QoS demands while maintaining manageable complexity through automated decision making
Data Source
AI summary
Systems, methods, apparatuses, and computer program products for beam selection using data radio bearer specific machine learning are provided. For example, a method can include providing one or more inputs regarding a plurality of beams to a machine learning model. The method can also include obtaining at least one output value regarding the plurality of beams from the machine learning model. The machine learning model can be a data radio bearer specific machine learning model, a data radio bearer group specific machine learning model, or a model trained to output selectively data radio bearer specific values or data radio bearer group specific values.