With the increase in the number of radar systems/radio applications, FMCW radars currently face various interference problems which lead to wider noise band, rise in noise floor level and masking of weak targets. This thesis aims to eliminate different types of interference for FMCW radars.
In this thesis, we propose a interference mitigation technique based on the signal
model. It is expected to detect the interferer at the beat frequency and mitigate the interference by reconstructing the interference-contaminated signal using the matrix pencil method in the time-frequency domain. However, there are two potential shortcomings of this approach that the interference needs to be detected and the loss of targets information is unavoidable. To solve the above problems, another model-based interference mitigation technique using matrix completion method is introduced. This method aims to decompose the measurement data to the desired signal and interference signal based on the low-rank property and the sparsity.
The method efficiency for different parameters of the interference, i.e. interference duration, signal-to-noise ratio (SNR) and different target scenarios ( i.e. a single stable target and distributed targets ) are investigated in numerical simulations. The proposed interference mitigation approaches also demonstrate experimental data collected by PASAX radar. Comparisons of the proposed approaches with the zeroing technique and other beat-frequencies interpolation techniques are performed. The results indicate that the interference can be significantly suppressed, with little cost in the accuracy of the reconstructed radar echo signals.