Optimizing peak age under intermittent satellite connectivity and store-and-forward

We consider a real-time task-oriented application operating over an intermittently available satellite-based communication network, aiming to collect status updates generated by a remote sensing device. The system is modeled as a scheduling problem over a finite horizon, corresponding to the duration of the task, to minimize the peak Age of Information at the destination. The number of updates that can be transmitted is constrained by a transmission budget. Moreover, the status updates are subject to delays caused by the store-and-forward operation of the satellites, which may vastly vary depending on the network conditions. We investigate three levels of awareness regarding the connectivity conditions of the satellite network: (i) scheduling without any information about connectivity conditions, (ii) scheduling based solely on the current conditions, and (iii) scheduling based on full connectivity knowledge. The first case admits a relatively simple structure, for which a periodic transmission strategy is adopted. The latter two cases are formulated as semi-Markov decision processes and solved to obtain the optimal transmission scheduling policy. Simulation results demonstrate the impact of connectivity awareness on the application performance at the destination. Through a simple modeling approach, we provide first insights into the practically relevant setting of store-and-forward satellite architectures.