Dynamic Neural Network Functions for 5G Positioning Measurement Data Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems, particularly in 5G networks, face challenges in efficiently processing positioning measurement data to achieve precise user equipment (UE) positioning due to limitations in spectral efficiency, signaling efficiency, and latency, especially with the increased complexity of multi-path propagation and beamforming in diverse environments.
Innovation Solution
The implementation of dynamically generated neural network functions based on machine-learning algorithms processes positioning measurement data into features, facilitating more accurate and efficient positioning by compressing data and filtering out irrelevant information, thereby enhancing positioning precision while managing signaling overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional positioning measurement data processing methods are used, then system complexity is low, but positioning accuracy is insufficient due to limitations in spectral efficiency and signaling efficiency
Solution Approach 1:
The patent introduces neural network functions as an intermediary component between the raw positioning measurement data and the positioning calculation process. These neural networks are deployed at the network side (LMF) and UE side, acting as mediators that automatically learn and extract relevant features from complex positioning measurement data, thereby improving positioning accuracy without requiring manual feature engineering and reducing the burden on traditional processing algorithms
Solution Approach 2:
The patent transforms the positioning measurement data through neural network processing, changing the parameters and representation of the data. The neural networks convert raw measurement data into optimized feature representations that are more suitable for positioning calculations, effectively changing the data parameters to achieve better positioning performance while managing computational complexity
2Measurement precision
If more positioning measurement data is processed, then positioning accuracy improves, but signaling overhead and latency increase
Solution Approach 1:
The patent performs preliminary processing of positioning measurement data using neural networks before the actual positioning calculation. The neural networks pre-extract and pre-process relevant features from the measurement data, so that when the positioning calculation is performed, the data is already in an optimized state, reducing the time required for final positioning computation and lowering overall processing latency
Solution Approach 2:
The patent extracts only the most relevant features from the positioning measurement data through neural network processing. Instead of processing all raw data, the neural networks identify and extract the critical features needed for accurate positioning, thereby reducing the amount of data that needs to be transmitted and processed, which in turn reduces signaling overhead and processing latency
3Measurement precision
If neural network functions are deployed for data processing, then positioning accuracy improves through adaptive learning, but device complexity and computational requirements increase
Solution Approach 1:
The patent segments the neural network processing into multiple components deployed at different locations: some neural network functions are deployed at the network side (LMF) while others are deployed at the UE side. This segmentation distributes the computational complexity across multiple devices, preventing any single device from becoming overly complex while still achieving the benefits of neural network processing for improved positioning accuracy
Solution Approach 2:
The patent designs the neural network functions to be multi-functional, serving multiple purposes: they process positioning measurement data, extract relevant features, adapt to different environmental conditions, and work with various types of positioning measurements (RTT, AOA, TOA). This universality allows a single neural network framework to handle diverse positioning scenarios, reducing the need for multiple specialized processing systems and thereby managing overall system complexity
Data Source
AI summary
In an aspect, a network component transmits, to a UE, at least one neural network function configured to facilitate processing of positioning measurement data into one or more positioning measurement features at the UE, the at least one neural network function being generated dynamically based on machine-learning associated with one or more historical measurement procedures. The UE may obtain positioning measurement data associated with the UE, and may process the positioning measurement data into a respective set of positioning measurement features based on the at least one neural network function. The UE may report the processed set of positioning measurement features to a network component, such as the BS or LMF.