WTRU Positioning Using AI Weighted Method Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication networks face limitations in positioning accuracy and latency due to reliance on specific measurement methods and centralized computation, which can be affected by environmental changes and availability of non-3GPP positioning methods.
Innovation Solution
Implementing artificial intelligence and machine learning techniques in wireless transmit/receive units (WTRUs) to autonomously determine the most accurate positioning methods by training neural networks with reference signal measurements and GNSS data, allowing for adaptive configuration of positioning reference signals and methods based on weight determination and preconfigured thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If centralized computation and specific measurement methods are used for positioning, then network control and standardization are maintained, but positioning accuracy and latency are limited
Solution Approach 1:
The patent divides the centralized positioning computation into distributed segments by enabling individual WTRUs to perform autonomous positioning computations using local neural network models. Each WTRU independently determines positioning methods and computes its own position, segmenting the previously centralized computation function across multiple distributed devices.
Solution Approach 2:
The patent implements self-service by enabling WTRUs to autonomously perform positioning computations using locally stored neural network models and configuration information. Each WTRU independently determines positioning methods, processes measurement data, and calculates position without requiring centralized network computation, allowing the system to serve itself.
2Adaptability or versatility
If multiple positioning methods are configured, then adaptability to environmental changes is improved, but device complexity and processing overhead increase
Solution Approach 1:
The patent implements dynamics by enabling WTRUs to dynamically select and switch between multiple positioning methods based on real-time environmental conditions and signal quality. The system transitions from static, pre-configured positioning methods to dynamic adaptation where the WTRU evaluates different methods and selects the most appropriate one for current conditions.
Solution Approach 2:
The patent applies parameter changes by modifying the weighting parameters of different positioning methods based on their performance and reliability. The neural network model adjusts the contribution of each positioning method through learned weight parameters, allowing the system to adapt to environmental changes by changing the parameters rather than the fundamental structure.
3Measurement precision
If WTRUs autonomously determine positioning methods using machine learning, then positioning accuracy is enhanced, but training time and resource consumption increase
Solution Approach 1:
The patent applies preliminary action by pre-training neural network models offline before deployment in WTRUs. The positioning methods and configuration information are determined and stored in advance, allowing the WTRU to immediately use the trained model for autonomous positioning without requiring real-time training, thus reducing operational training time.
Solution Approach 2:
The patent uses simplified, lightweight neural network models that can be quickly trained and deployed. Rather than using complex, time-consuming training procedures, the system employs efficient model architectures that require minimal training resources and time, making the training process disposable and replaceable if needed.
4Reliability
If reference signal measurements and GNSS data are used for training, then positioning reliability is improved, but dependency on external signals increases
Solution Approach 1:
The patent merges multiple positioning data sources including reference signal measurements, GNSS data, and other available positioning methods into a unified neural network model. By combining these diverse inputs, the system creates a more robust positioning solution that can cross-validate and compensate for weaknesses in individual signal sources.
Solution Approach 2:
The patent dynamically changes the weighting parameters of different signal sources based on their availability and quality. The neural network model adjusts the contribution of reference signals, GNSS data, and other methods according to current conditions, reducing dependency on any single external signal while maintaining overall reliability.
Data Source
AI summary
The disclosure pertains to methods and apparatus for using artificial intelligence and machine learning for positioning of nodes (e.g., wireless transmit/receive units (WTRUs)) in wireless communications. In an example, a method implemented by a WTRU for wireless communications includes receiving configuration information indicating a plurality of positioning methods and a threshold, determining a respective weight for each of the plurality of positioning methods, and sending the respective weights for the plurality of positioning methods based on determining that at least one of the respective weights is greater than the threshold and/or after a preconfigured time period.