University of Twente Student Theses


Direction-of-Arrival Estimation Using A Machine Learning Framework

Hijlkema, Frank (2022) Direction-of-Arrival Estimation Using A Machine Learning Framework.

[img] PDF
Abstract:In recent years, radar technology has emerged in various civil applications, and one of them are the automotive vehicles. Modern cars have several radars that are used to improve driver experience for instance in providing parking assistance. These radar sensors are placed at different positions on the vehicle to gather information about object around the car. The echoed signal from the objects are impinging on a Minimum Redundancy Antenna (MRA) which is used to estimate the Direction of Arrivals (DoAs). The DoA is calculated using Maximum Likelihood Estimation (MLE) which is considered as the state of the art algorithm for MRAs. However, the high computational complexity and the imperfections that are present in antenna arrays makes this approach computationally expensive. This thesis investigates the opportunities of using machine learning for DoA estimation in these automotive radar systems. We have focussed on different aspects of the DoA estimation problem for instance imperfections, combined azimuth and elevation estimation, cost minimization and number of targets estimation. These aspects have been treated as classification problems and were considered only for fully connected neural networks. The DoA estimation results obtained with our neural networks are compared with the results obtained with MLE. Our results have indicated that machine learning can be an attractive approach for the existing challenges of traditional algorithms. Using this approach, the impact of imperfections is mitigated compared to traditional algorithms. Furthermore, making use of machine learning can decrease the required computational load by a factor of 200 to a modest memory usage of 256 kB. This reduction is obtained through combining a shallow neural network with MLE which makes it attractive to apply neural networks in automotive radar systems.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Embedded Systems MSc (60331)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page