Browsing by Author "Alexander, Faustmann"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Polyspectral methods as tools for uncovering the statistics of pulsar radio signals.
    (Stellenbosch : Stellenbosch University, 2023-03) Alexander, Faustmann; Gilmore, Jacki; Schwardt, Ludwig; Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering.
    ENGLISH ABSTRACT: Although the first r adio p ulsar w as d iscovered 5 5 y ears a go, t here i s s till a l ack o f consensus concerning the mechanism responsible for the observed radio emissions. Additionally, the improvement in observing instrumentation has achieved a poorer improvement in timing precision than anticipated. The number of catalogued pulsars to date is also far below the theoretical and observable galactic population. These issues have motivated this study, which applies novel signal processing techniques to pulsar radio signals. The higher-order stationary moment spectra are the Fourier companions of the higher-order stationary moment functions. They offer a host o f p owerful statistical tools and are useful for system identification a nd n on-Gaussian p rocess c haracterization. T he p olyspectra g eneralize t he stationary moment spectra to the non-stationary case and are intimately linked to correlations of spectral processes. The noise spectrum comprises the variances and covariances between channels of the power spectrum; it is related to the polyspectrum and its inverse Fourier transform yields the estimation error of the autocorrelation function. This estimation-error function is able to discriminate an intermittent process from a Gaussian one and also to estimate the statistics of the intermittency. The work presented here applies methods based on the higher-order spectra, polyspectra, and noise spectra to pulsar radio signals. The experiments performed over the course of the investigation are unique in the field of radio pulsar a stronomy. The primary contribution of the work is the extension of the noise spectra technique to a Bernoulli-sampled Gaussian process and the application of the method to radio signals from six pulsars. The final result i s evidence in support o f a signal model for pulsar radio signals that is sparse and impulsive. If this is indeed the case, then it explains the excess timing noise, provides observational evidence for constraining the emission mechanism and also provides a possible feature for advanced detection algorithms.