AI/ML Model Pairing for Entity-Specific Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless networks face challenges in optimizing network operations and end device performance due to the need for entity-specific configurations and optimizations that are not adequately addressed by standardized AI/ML models, leading to suboptimal Quality of Service (QoS) and Quality of Experience (QoE).
Innovation Solution
A learning-based network optimization service utilizing entity-specific AI/ML models and model pairing services to tailor network operations and end device capabilities, incorporating RF modem identifiers and chipset information for customized optimizations and performance enhancements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If standardized AI/ML models are used for network optimization, then implementation complexity is reduced, but QoS and QoE performance deteriorates due to lack of entity-specific optimizations
Solution Approach 1:
The patent segments the AI/ML model into two distinct components: a standardized base model that handles general network optimization functions, and an entity-specific model that tailors optimizations to specific end devices based on their capabilities (RF modem identifiers, chipset information). This segmentation allows the system to maintain low implementation complexity through the standardized base while achieving high QoS/QoE performance through device-specific customization.
Solution Approach 2:
The patent applies local quality by adapting the AI/ML model to specific local conditions of each end device. The system collects device-specific information (RF modem identifiers, chipset information) and uses this to customize the model parameters and configurations for each device, ensuring optimal performance tailored to local device characteristics rather than applying a one-size-fits-all approach.
2Reliability
If entity-specific AI/ML models are implemented, then QoS and QoE performance is improved, but device complexity and LCM process complexity increase
Solution Approach 1:
The patent implements preliminary action by pre-collecting and storing entity-specific information (RF modem identifiers, chipset information) during device initialization and registration phases. This advance preparation allows the system to quickly retrieve and apply appropriate model configurations during operation without adding complexity to real-time model management, as the device characteristics are already known and stored for future reference.
Solution Approach 2:
The patent introduces an intermediary component (the network device that manages AI/ML models) that acts as a mediator between the standardized model repository and the end devices. This intermediary handles the complexity of model selection, configuration, and updates by translating device-specific requirements into appropriate model parameters, thereby shielding the overall system from the full complexity of managing numerous entity-specific models while still delivering customized performance.
3Ease of operation
If generic network optimization approaches are used, then ease of operation is maintained, but network and end device performance deteriorates
Solution Approach 1:
The patent implements self-service by enabling the system to automatically collect device-specific information (RF modem identifiers, chipset information), select appropriate model configurations, and apply optimizations without requiring manual intervention. The network device autonomously manages the entire process of retrieving device characteristics, matching them with suitable model parameters, and implementing customized optimizations, thereby maintaining ease of operation while achieving high performance through entity-specific tailoring.
Data Source
AI summary
A method, an end device, and a non-transitory computer-readable storage medium are described in relation to a learning-based network optimization service. The learning-based network optimization service may include an AI/ML model pairing between an end device and a network device based on AI/ML model identifiers that may include or correlate to a radio frequency (RF) modem or chipset identifier of the end device. The AI/ML models may support entity-specific optimizations, use and sub-use cases, and network-side and end device-side key performance indicators.