- Dorschky, Eva
Transient detection, another key research objective of the SKA, is the focus of this work. The current detection method estimates the autocorrelation of the observation data for every relevant dispersion measurement (DM) coefficient. The wide range of the DM coefficients, however, causes an exponential computational effort. The detection of astronomical transients can be shown to be an NP-hard problem.
In this thesis, a computationally efficient algorithm for transient detection in astronomical data shall be investigated based on cluster analysis. The work plan includes the statistical modeling of pulsars in Matlab, the clustering of modelled observation data via the implementation of an EM-algorithm (including convergence tests) and transient isolation/identification. The clustering process groups together observations which are likely to be drawn from the same probability distribution. One cluster group represents the observations that are likely to be just noise. The isolation/identification of the transients is performed on the signal cluster group(s) via a curve-fitting autoregression method. Simulations are to be carried out to verify the detection capability of the modelled pulsars. Finally, the complexity of the clustering approach versus full correlation techniques is to be analyzed.