ARGUS_JSA_Supplimentaryfile

Description

ARGUS: Pricing Adaptation-Induced Security Risk in Zero-Trust Edge Video Analytics”, as an original research article. Adaptive edge video analytics is judged in service on two things: whether the deadline is met, and whether the analytics can be trusted when it is. Orchestrators re-tune resolution, model, split point and offload tier to hold latency and energy budgets, and every one of those knobs also moves adversarial robustness. No orchestrator in the literature prices that movement, so the quality loop manufactures security risk that nothing in the system measures, while zero-trust verification sits outside the loop and consumes the same latency budget the orchestrator is trying to protect.

Categories

Licence