Cellular Network Geo-Location Analysis Using Unsupervised Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing geo-location data in cellular networks are limited in accurately identifying areas with network issues, such as dropped calls and poor coverage, and do not effectively utilize unsupervised learning to generate transformation functions for optimizing network performance.
Innovation Solution
A method and system that processes data using a processor to generate geo-locating transformation functions through unsupervised learning, allowing for the geo-location of mobile devices and the calculation of metrics like dropped calls and network quality, by associating track points with geo-coordinates and time-stamped data, and updating these functions based on Self-Organizing Network (SON) corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to collect and analyze geo-location data, then data collection is simple, but measurement precision and accuracy of network issue identification deteriorate
Solution Approach 1:
The system performs preliminary actions by collecting extensive raw network data and device information in advance, storing it in databases for later processing. This preliminary data collection phase enables more accurate measurements and analysis when needed, without requiring complex real-time processing during network operations.
Solution Approach 2:
The patent introduces intermediary components including data processing modules, analysis engines, and transformation functions that act as mediators between raw data collection and final network issue identification. These intermediaries process and transform data systematically, improving measurement precision while managing system complexity through modular architecture.
2Measurement precision
If detailed geo-location tracking is implemented, then network quality measurement accuracy improves, but loss of time and processing overhead increase
Solution Approach 1:
The system collects and pre-processes geo-location data, device information, and network parameters in advance, storing them in organized databases. This preliminary action reduces the time required for actual network quality analysis by having data ready for rapid processing when measurements are needed.
Solution Approach 2:
The patent implements periodic data collection and processing cycles, where geo-location data and network parameters are gathered at regular intervals rather than continuously. This periodic approach maintains measurement accuracy while reducing overall processing time and computational overhead compared to continuous tracking.
3Adaptability or versatility
If unsupervised learning transformation functions are generated, then adaptability and network optimization improve, but device complexity and computational requirements increase
Solution Approach 1:
The system employs self-service mechanisms through unsupervised learning transformation functions that automatically analyze collected data and generate network optimization insights without requiring manual intervention or external supervision. The learning system serves itself by autonomously identifying patterns, correlations, and optimization opportunities from the accumulated geo-location and network data.
Solution Approach 2:
The patent performs preliminary data organization, cleaning, and structuring before applying unsupervised learning algorithms. This preliminary preparation of data reduces the computational complexity of the learning process by presenting organized, high-quality input data to the transformation functions, making the system more manageable despite the inherent complexity of machine learning.
4Reliability
If comprehensive data collection is performed, then reliability of network analysis improves, but loss of information and data management complexity increase
Solution Approach 1:
The patent segments comprehensive network data into distinct categories and components, including geo-location data, device information, network parameters, and performance metrics. This segmentation organizes the comprehensive data collection into manageable segments that can be processed and analyzed systematically, reducing data management overhead while maintaining the reliability benefits of comprehensive collection.
Data Source
AI summary
There is provided a method of analyzing data collected in a cellular network and system thereof. The method comprises: repeatedly processing a first subset of the collected data to geo-locate mobile devices, thereby giving rise to geo-location data informative of geo-located mobile devices and associated time-stamped geo-locations thereof, wherein geo-locating is provided with the help of a geo-locating transformation function generated by unsupervised learning, by the processor, of a second subset of the collected data. The method further comprises processing the geo-location data to obtain time-stamped track points, each track point is associated with a respective mobile device and is characterized by geo-location, azimuth and velocity; storing the obtained track points in association with data indicative of respective mobile devices and time-stamps, thereby generating a geo-location database; and processing the geo-location database, to obtain one or more metrics of the cellular network as a function of, at least, respective geo-coordinates.