For more than ten years, the Society of Aeronautics and Astronautics has awarded scholarships to the best degree project of the year associated with aeronautics and astronautics at Chalmers. Among the four nominated master's theses 2019, one stood out both in terms of the content and the design of the report; Erik Wallin's thesis ”Detecting Jamming and Interference in Airborne Radar Using Convolutional Neural Networks”.
“I am very happy that my master’s thesis was appreciated”, says Erik Wallin, who received extra praise from the committee for his way of writing the report, allowing also people not familiar with the subject to understand it.
In his degree project, he investigated how machine learning can be used as a tool to classify and distinguish different types of disturbances and interferences, that an airborne radar system can be exposed to.
“Modern flight radar systems have to endure exposure to both analogue interference noise and digital interference attacks”, Erik explains. “Digital interference is more difficult to detect and manage, as it can be hidden in other signals. Using simulated data, I trained a neural network to distinguish and classify different types of interferences.
The task was passed with flying colours. As stated in the justification of the award: “The results show that future radar systems can be designed considerably more robust against most types of interference. An effective classification of interference in radar data is also an important step towards a cognitive radar, a highly topical area of research, which means a radar that autonomously adapts its behavior to the current environment.”
From degree project to an employment contract
Erik studied Engineering Physics at Chalmers with a master’s degree in Complex Adaptive Systems. After graduation, he immediately got a job at Saab Surveillance in Kallebäck.
“My years as a student at Chalmers have laid a good foundation for my working life”, he says. “Especially, I have benefited from the courses I took in math and physics. Radar technology was new to me, so I got to know it step by step at Saab.
Now, the next challenge is awaiting him. Erik just got a new position as an industrial doctoral student in the research group Signal processing. He will spend 80 percent of his time at Chalmers and the rest at Saab as a part of the research program WASP (Wallenberg Artificial Intelligence, Autonomous Systems and Software Program).
“This is an opportunity I don’t want to miss”, he says. “The degree project gave me the insight that I enjoy exploring a problem by myself that no one else has solved before. At the same time, I will get the privilege of learning from the academic setting during my five years of doctoral studies.”
Is it a bird or a drone?
One problem Erik is about to tackle now is how to teach sensor systems to distinguish between flying birds and drones.
“I will use artificial intelligence in so-called semi-supervised learning, where both labeled and unlabeled data are used for training a computer program to perform the task. Usually this is done by analysing large amounts of images, but in this case I will use radar data instead. That’s something there is plenty of at Saab!”
Text and photo: Yvonne Jonsson
More about the awarded master’s thesis
The project was performed at Saab Surveillance in Gothenburg. Supervisor at Saab was Albert Nummelin, and examiner at Chalmers was Thomas Rylander, Professor at the department of Electrical Engineering.