UAV Detection via Machine Learning Classification in Wireless Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Service providers face challenges in distinguishing Unmanned Aerial Vehicles (UAVs) from other user equipment devices in wireless networks, which hinders effective network management and resource allocation, impacting Quality of Service (QoS) and service level agreements (SLAs).
Innovation Solution
A machine learning model using logistic regression is employed to classify UAV-related records from non-UAV records based on connection-related data, with continuous updates using data from known UAVs to enhance accuracy, enabling the identification of UAVs and optimizing network resource management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional network management methods are used to treat all user equipment uniformly, then network operations are simple, but UAV-specific QoS requirements cannot be met
Solution Approach 1:
The patent segments user equipment into different categories (UAVs and non-UAVs) using machine learning classification. By analyzing connection-related data patterns, the system identifies UAV-specific behaviors and separates them from general UE traffic, enabling differentiated QoS policies for each segment while maintaining overall network management efficiency
Solution Approach 2:
The patent introduces an intermediary machine learning model that acts as a mediator between raw network data and QoS decision-making. This intermediary classifies UEs based on connection patterns and provides structured output to network management systems, simplifying the integration of UAV detection capabilities without requiring complex changes to existing network infrastructure
2Measurement precision
If machine learning models are continuously updated with new data, then classification accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary actions by pre-processing and storing connection-related data in structured formats during normal network operations. Training data is prepared and organized in advance, allowing the machine learning model to be trained efficiently when updates are needed, reducing the time penalty associated with continuous learning
Solution Approach 2:
The patent implements continuous learning by periodically updating the machine learning model with new connection data while maintaining model persistence. Instead of complete retraining, the system continuously refines classification accuracy by incorporating new patterns, ensuring the model stays current with evolving UAV behaviors without excessive computational overhead
Data Source
AI summary
A method may include obtaining connection-related data associated with user equipment (UE) devices operating in a network, wherein the UE devices include unmanned aerial vehicles (UAVs) and devices other than UAVs. The method may include storing the connection-related data; identifying, in the stored connection-related data, data associated with known UAVs and filtering the stored connection-related data based on a distance associated with a known UAV. The method may further include training a machine learning classifier using the filtered data and executing the machine learning classifier to identity UAVs operating in the network.