Exogenous Update Scheduling in the Industrial Internet of Things for Minimal Age of Information

Kurzfassung

Many studies about real-time applications quantify data freshness through age of information (AoI), and assume that fresh data can be generated at any time. However, industrial Internet of things (IIoT) applications for alert, monitoring, or task-oriented activities follow an exogenous data generation and often operate over a finite time window. This motivates the present analysis, where we study the AoI-minimizing scheduling of status updates for an IIoT source generating fresh data only at random instants. We distinguish between infinite and finite horizon, the latter being more consistent with IIoT tasks. For each case, we discuss either agnostic (predefined and unmodifiable) or source-aware scheduling, depending on the probability of fresh data generation and duty cycle. We give a tight bound for source-aware infinite-horizon scheduling and exact expressions for the other three cases. We evaluate the surge in AoI due to sporadic generation, with worst-case increases of 3 and 2 times for agnostic and source-aware scheduling, respectively. However, these are pessimistic estimates when the probability of data generation is at least one order of magnitude higher than the duty cycle, in which case the increase is less than 20% for agnostic scheduling, and almost negligible in the source-aware case.