Quota-Guided Deadline Protection and AoI-Aware Scheduling for Mixed-Criticality Event Streams
27th International Conference on Computer Systems and Technologies, CompSysTech 2026, Hybrid, Ruse, Bulgaristan, 26 - 27 Haziran 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/compsystech69798.2026.11683078
- Basıldığı Şehir: Hybrid, Ruse
- Basıldığı Ülke: Bulgaristan
- Anahtar Kelimeler: edge computing, mixed-criticality, real-time scheduling, stream processing
- Atatürk Üniversitesi Adresli: Evet
Özet
Real-time Information Technologies (IT) systems often process heterogeneous event streams over the same limited computing resource. In edge nodes, stream-processing operators, cyber-physical monitoring platforms, and intelligent transportation services, freshness-critical sensor updates, strict-deadline warning messages, and best-effort background traffic may arrive at the same processing unit. This creates a reliability-freshness conflict: deadline-critical messages must be served before they expire, while sensor updates must still be delivered often enough to keep the available information recent.This paper presents a quota-guided scheduling framework for mixed-criticality event streams. The framework decides which packet should be served next by combining three mechanisms: guard-based deadline protection, quota-based service regulation, and Age-of-Information-aware update selection. Deadline alerts are prioritized when their remaining time becomes critical. When they are not urgent, the quota mechanism limits long consecutive service periods of deadline traffic and preserves service opportunities for freshness-critical updates. Age of Information (AoI) is used to select sensor updates according to information freshness.The framework is evaluated using a discrete-event simulation model with three traffic classes: freshness updates, deadline alerts, and best-effort background traffic. The proposed policies are compared with First-In-First-Out (FIFO), Earliest Deadline First (EDF), and AoI-Max under light, medium, and heavy load conditions. The results show that the proposed mechanisms keep deadline-alert violation ratios at zero or below 0.5% across the tested load regimes, while maintaining bounded AoI tail behavior. In contrast, the baseline policies either degrade deadline reliability or produce larger freshness degradation under congestion.Parameter-sensitivity results further show that the proposed scheduler remains stable over a broad range of quota, guard, and gap settings. These findings indicate that quota-guided deadline protection provides a lightweight and configurable scheduling layer for edge computing, stream processing, cyber-physical monitoring, and other real-time IT infrastructures.