| Summary: | This master thesis has been performed at SAAB AB in Järfälla, Sweden.A radar warning receiver must alert the user when someone highlights it with radarsignals. Radar signals used today varies and has a wide frequency band. In order todetect all possible radar signals the radar warning receiver must have a widebandwidth. This results in that the noise power will be high in the radar warningreceiver and weak radar signals will be hard to detect or even undetected.The aim of the thesis work was to investigate the possibility to suppress the noise inthe received radar signals. Unfortunately we do not know the frequency of thereceived radar signals, since the frequency has been decided by the threat radar. Wehave used adaptive filters, which adapts it band-pass to the received radar signal. Theadaptive filters must converge quickly to the state it reduces the noise and passes theradar signals since radar pulses can be very short in the time domain. We also wantto achieve a high SNR gain that is a measurement of how well the adaptive filterreduces the noise.We have investigated two adaptive algorithms, the recursive least square (RLS)algorithm and the least mean square (LMS) algorithm. We found out that the LMSalgorithm was more suitable for noise cancellation in radar applications due to its lowcomplexity and stability compared to RLS algorithm. The LMS algorithm gave SNRgains in the span 14-20 dB for different radar pulses.
|
|---|
