For my master’s thesis, I have a lot of data of state of polarization of light (particularly S1 parameter) from undersea fibre optic cables. Now I want to perform anomaly detection on this data so that I can flag any external events like fibre tampering, nearby vessels or earthquakes. The data is unlabelled and I don’t have any idea what an anomaly can look like. So I decided to preprocess the signals and generate the typical FFT, STFT plots. The following plot is obtained for one file (~60 seconds of data sampled at 44.1KHz). I used the log scale to see the entire frequency distribution properly. Any tips on what to infer from this data and how to move forward would be much appreciated. PS: I am completely new to signal processing domain.For my master’s thesis, I have a lot of data of state of polarization of light (particularly S1 parameter) from undersea fibre optic cables. Now I want to perform anomaly detection on this data so that I can flag any external events like fibre tampering, nearby vessels or earthquakes. The data is unlabelled and I don’t have any idea what an anomaly can look like. So I decided to preprocess the signals and generate the typical FFT, STFT plots. The following plot is obtained for one file (~60 seconds of data sampled at 44.1KHz). I used the log scale to see the entire frequency distribution properly. Any tips on what to infer from this data and how to move forward would be much appreciated. PS: I am completely new to signal processing domain. For my master’s thesis, I have a lot of data of state of polarization of light (particularly S1 parameter) from undersea fibre optic cables. Now I want to perform anomaly detection on this data so that I can flag any external events like fibre tampering, nearby vessels or earthquakes. The data is unlabelled and I don’t have any idea what an anomaly can look like. So I decided to preprocess the signals and generate the typical FFT, STFT plots. The following plot is obtained for one file (~60 seconds of data sampled at 44.1KHz). I used the log scale to see the entire frequency distribution properly. Any tips on what to infer from this data and how to move forward would be much appreciated. PS: I am completely new to signal processing domain. fft, signal processing, digital signal processing, signal, matlab MATLAB Answers — New Questions
