Scalable runtime monitoring of radio telescope signals

Abstract

In the past few decades, new discoveries about the universe, including pulsars, have led the way to more understanding of how the universe has evolved. Pulsars are neutron stars which are rapidly rotating and emitting electromagnetic radiation beams. These electromagnetic signals are detected by radio telescopes, which detect radio signals from the sky. The output of such telescopes which takes the form of time-sampled data is be used, with the main aim of capturing and detecting these pulsars. To carry out this task, we use a technique called runtime verification. It is a method which verifies the correctness of software by pattern matching the behaviour of the execution to detect bugs. The same idea is applied here, but instead of bug detection, identification of pulsars by pattern recognition is performed. The advantage of this approach is that the time-sampled data is gathered and analysed progressively unlike the current detection techniques where a buffer is used to process the data. We gain a performance of O(N), having the execution time proportional to the number of samples. Consequently, our solution displays temporarily results after the first period detected. Results are promising and the technique can be used in real scenarios.