Crowd-Sourced Base Station Location Estimation Using ML Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional drive testing for determining base station locations is limited in scope and becomes outdated quickly, while crowd-sourced data from mobile devices can provide real-time information but includes inaccuracies that lead to erroneous cell site location estimation.
Innovation Solution
Employing machine learning models, such as clustering models like DBSCAN, to filter out inaccurate data from crowd-sourced information before using algorithms like trilateration to accurately determine base station locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If crowd-sourced data from mobile devices is used to estimate base station locations, then real-time information is provided, but inaccuracies in the data lead to erroneous location estimation
Solution Approach 1:
A machine learning model acts as an intermediary between the raw crowd-sourced data and the location estimation algorithm. The model processes the noisy input data, identifies patterns, and filters out anomalies before the cleaned data is used for trilateration, thereby resolving the contradiction between using real-time data and maintaining accuracy
Solution Approach 2:
The patent replaces traditional mechanical filtering methods with machine learning-based anomaly detection. Instead of using simple threshold-based filtering or geometric validation, the system employs trained models to intelligently distinguish between valid and erroneous measurements, achieving both real-time processing and high accuracy
2Measurement precision
If conventional drive testing is used to determine base station locations, then accurate location data is obtained, but the scope is limited and data becomes outdated quickly
Solution Approach 1:
The system transforms mobile devices from single-purpose communication tools into multi-functional sensing platforms. By utilizing the GPS, cellular radio, and processing capabilities of ordinary smartphones for location estimation, the system achieves both wide geographic coverage and continuous data collection without requiring specialized drive testing equipment
Solution Approach 2:
The patent enables crowd-sourced data collection where mobile devices automatically contribute their location and signal measurement data without requiring manual intervention. The devices serve themselves as both sensors and processors, continuously generating location estimation data that feeds into the machine learning model, thereby achieving extensive coverage and real-time updates
Data Source
AI summary
Crowd sourced data from mobile devices may be used to estimate site locations. In addition, machine learning models may be used to filter out any inaccurate crowd sourced data before using algorithms to estimate the cell site location. An apparatus may include a processor and a memory coupled with the processor that effectuates operations. The operations may include receiving data associated with a location of plurality of devices that have connected with a base station in a geographic area; determining a machine learning model to apply to the received data based on the type of data, wherein the machine learning model is a clustering model; performing the clustering model on the received data; based on the performing the clustering model on the received data, obtaining representative data that excludes outliers in the received data; based on the representative data, determining a location of the base station in the geographic area; and sending a message with the location of the base station.