Examenspresentation av Usama Abdullah och Zubia Mumtaz

Examiner: Tomas McKelvey

Abstract:
In this thesis, performance metrics are established to evaluate the efficiency and accuracy of various grid-less multiple target detection algorithms in radar. A high-fidelity radar simulation environment is established with the liberty to change parameters affecting the radar base-band frequency estimation. The radar data is generated using a base-band sinusoidal signal model, in which clutter and noise can be added or removed. Different settings of data generation are used to replicate multiple realistic scenarios. The detection algorithms are rigorously tested under these scenarios including high noise and dense clutter environment. Special cases when two target frequencies are very close to each other are also monitored. The errors are recorded for various noise levels, signal lengths, and clutter spread and strengths. A mapping metric i.e. GOSPA, is used to correlate estimated and actual parameters to perform an effective evaluation. This work aims to provide a detailed evaluation platform for detection algorithms. Moreover, different clutter removal techniques are also introduced for the detection algorithms.

Welcome!

Usama, Zubia and Tomas