New Method Ensures Real-Time IoT Data Freshness Safety

This paper addresses a critical safety challenge in IoT systems, particularly those involved in industrial control, V2X communication, and remote teleoperation: ensuring that the "peak Age of Information" (PAoI) for every sensor remains below a strict, per-slot deadline. Existing solutions often rely on restrictive assumptions or only guarantee average bounds, which are insufficient for safety-critical applications. The researchers propose OCO-PAoI-Hard, a new method that guarantees zero per-slot violations of the modeled AoI state, even when faced with adversarial coefficients. The core insight is to transform the fractional peak-AoI deadline into an affine half-space constraint on the resource-allocation vector, effectively turning hard real-time scheduling into a time-varying constrained online convex optimization problem over a polyhedral safe set. The system employs a strictly causal proposal-shield-update loop, enforcing feasibility with a single Euclidean projection per slot, while a gradient step maintains no-regret behavior. Evaluated on a four-sensor adversarial fluid-model trap channel, OCO-PAoI-Hard achieved zero modeled-state deadline violations across all test seeds, significantly outperforming four representative baselines that missed between 1.65% and 64.0% of slots.