| Title |
Radijo siųstuvų klasifikavimo, naudojant Hilberto-Huango transformacijos metodą ir neuroninius tinklus, tyrimas / |
| Translation of Title |
Research of radio transmitter classification using Hilbert-Huang transform method and neural networks. |
| Authors |
Paulauskas, Aretas |
| Full Text |
|
| Pages |
99 |
| Keywords [eng] |
Hilbert-Huang transformation spectrum ; transient signals ; digital signal processing ; deep learning |
| Abstract [eng] |
The main topic is to devise an algorithm to classify various radio transmitters based on their transient signals. In the first part, a review is made on a diverse set of other researchers work, used algorithms and calculations. A comparison is made between machine and deep learning – a short investigation is also done on how various extracted data features impact the classification performance of trained networks. Also, a transient signal detection algorithms are reviewed as well as various ways to analyze a complex signal. The second part consists of data collection system as well as equipment used in the process. Simplified diagrams are given and the used equipment is described. The third part consists of the captured data analysis – various modulation schemes were analysed, including FSK, GFSK and LoRa. The analysis was made using Hilbert-Huang transform method and compared to Short – Time Fourier Transform. Data augmentation technique is used to increase the training data and make the trained network more robust. The fourth part is where neural network classification takes place. A comparison between three different networks is made: ResNet-34, VGG-16 and a CNN network taken from analyzed authors, which was created on ResNet basis. |
| Dissertation Institution |
Kauno technologijos universitetas. |
| Type |
Master thesis |
| Language |
Lithuanian |
| Publication date |
2024 |
