Dispatch approaches for scheduling radio telescope observations

Abstract

Radio telescopes are a scarce resource designed to provide experimental data for astrophysical research, and various studies have focused on the design of observing schedules that make optimal use of available telescope time. We consider strategies for minimising the time required to observe a fixed set of pulsars, which provides an excellent example of scheduling in an unpredictable environment. First, owing to scintillation, the intensity of a pulsar signal is variable and random; therefore, the decision to abort or prolong an observation can be made only after some fraction of the scheduled observation has been completed. Second, observations may be interrupted by radio frequency interference or when the source sets below the horizon. Some sources are visible for more or less time depending on their declination and the latitude of the observing telescope. Formulating the problem in these terms leads to a highly dynamic shortest path problem with uncertainty. Unlike other documented telescope scheduling approaches, we demonstrate how a simple earliest setting policy achieves sets of pulsar observations in a rather short timespan. The policy is fast to apply due to a novel algorithm that pre-calculates the subset of next candidates before the end of the current integration. Our simulation also clarifies that the uncertainty arising from the scintillation (signal strength) encountered on arrival adds significantly to the variation in overall durations and that different hourly start times can favour or hamper the progress of a set of observations.