| The thesis proposes a matched filter approach to detect quadrupole resonance (QR) signals in the presence of disturbance from AM stations. Detecting QR signals is a challenge due to several reasons. One is the amplitude of a QR signal is typically on the same level of thermal noise, which makes it very susceptible to noise interferences. External Radio Frequency (RF) interferences, such as AM signals, and internal RF interferences, ones from inside the search volume, pose another challenge and contribute to the low SNR values observed. AM stations broadcast within the same frequency band of QR signals, which is a problem for QR detection. A third important challenge we face is the uncertainty in the QR signal characteristics.;To motivate the use of a matched filter approach, a matched filter (under the assumption that the QR signal is known) was compared to the generic energy detector in theory and it resulted in a performance improvement. The work proposes a detector referred to as the batch matched filter, which uses a gridding technique to search for unknown QR signal parameters and attempts to match the filter to the shape of the QR signal present. This approach resulted in a performance gain when compared to the generic energy detector using simulation and experimental data, where the QR signal is unknown. To further improve performance we introduced an approach that would also match the filter to the noise present in addition to the QR signal. This approach is referred to as the batch whitened matched filter and when properly matched to the noise outperforms both the batch matched filter and energy detector. |
