• Home
• Graduate
• Future Students
• Research
• Potential Research Topics
• Machine Learning Meteor Echo Detection

Application of Machine Learning to Radar Meteor Echo Detection

Supervisor: Dr. Peter Brown

Project Description (Abstract):

This project involves development of a convolutional neural network (CNN) to automatically detect and classify the signatures of different types of meteor echoes in raw radar data. Data from the Canadian Meteor Orbit Radar (CMOR) will be used to train the CNN and extract signal features of interest. Once automated detection is achieved, algorithms will be developed building on existing software packages to extract signal metrics from each type of echo.

The final outcome of the project will be an operational CNN which will extract meteor echo features constantly from the real-time radar stream and perform basic measurements.

The student will use existing code written in C to extract basic data from the raw radar data. Development of CNN test code in Python or C will be used to construct a CNN for the detection and classification of meteor radar echoes.