New approaches for the real-time detection of binary pulsars with the Square Kilometre Array (SKA)

Data rates from the Square Kilometre Array will be huge, rivalling the sum total of current global internet traffic. This deluge of data prompts a demand for significant progress in techniques for signal processing, data analysis and time-series modelling of vast data sets. Developing novel, real-time, machine learning algorithms is paramount if the SKA project is to meet its science objectives [1]. As a case study in SKA data analysis, this paper outlines some intrinsic difficulties in the search for binary pulsars, briefly describes known techniques for binary pulsar detection and proposes new detection approaches. Our current focus is to model more accurately sources of non-Gaussian, non-stationary noise in a typical pulsar search. Any solution must be scalable, to satisfy the real-time requirement of the SKA pulsar surveys.