Answers are grounded in this paper and cite their sources.
Cyber-physical systems such as UAV swarms increasingly face cross-layer attacks in which adversaries weaponize cryptographic workloads to induce scheduling delays, CPU starvation, and physical instability. This paper develops a framework that co-designs input-to-state stability (ISS) with the symmetric and asymmetric processing overheads of the cryptographic stack. Cryptographic verification delays and adversarial computational loads are modeled as endogenous, state-dependent terms within a switched delay system, enabling Lyapunov–Krasovskii stability guarantees under resource-constrained operation. Exploiting the cost asymmetry between cryptographic primitives, the framework transitions between lightweight symmetric verification during nominal operation and stronger asymmetric verification under detected anomalies, with the escalation certified by the same functional that establishes stability. To enforce the certified delay bound by load shedding, the framework employs admission control, dropping excess packets at the network interface when the incoming workload exceeds CPU capacity. The empirical evaluation uses 33 measured attack captures from a publicly available UAV network intrusion detection dataset, spanning 13 attack classes with durations from 18.4 s to 3121.7 s. Converting measured packet rates into CPU load shows that eight of the 13 classes exceed unity load, with distributed denial of service (DDoS) reaching approximately 20× saturation and DoS reaching approximately 10×. Under a common attack schedule generated from this distribution, an unprotected verification path diverges while the proposed mechanism remains bounded by capping the queue at the certificate limit; a bandwidth sweep on the canonical plant identifies approximately 8 rad/s as the threshold above which the unprotected system destabilizes.
No takes yet. Share an insight, caveat, or question.
Ohemeng et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: