Work Packages

The SPIREC project is organized into four complementary work packages, each addressing specific aspects of multi-level supervision and prediction for geo-distributed and heterogeneous infrastructures.

Work Package 1

Data Models and Observability

• Objective: Establish foundational data representations and monitoring frameworks for the Cloud-Edge-IoT continuum.
• Scope: Focuses on defining SLA/SLO specifications for highly distributed environments, modeling metrics to infer causal dependencies without massive datasets, and designing lightweight observability architectures tailored for multi-access edge infrastructures.

Work Package 2

Supervision and Anomaly Detection

• Objective: Detect, isolate, and troubleshoot system failures across heterogeneous environments.
• Scope: Investigates low-layer (Edge/IoT) anomaly detection under tight resource constraints. It develops advanced Root Cause Analysis (RCA) methods to trace and isolate complex, non-linear failures within distributed microservice architectures.

Work Package 3

Resource Prediction and Distributed AI

• Objective: Forecast system load and optimize collaborative learning frameworks across the continuum.
• Scope: Centers on developing distributed machine learning and time-series prediction models. The focus is on forecasting traffic and resource consumption while designing hierarchical federated learning architectures that operate within local resource bounds and protect global SLOs.

Work Package 4

Tools, Integration, and Validation

• Objective: Coordinate technical integration, software deployment, and large-scale validation.
• Scope: Acts as the practical implementation engine by integrating software prototypes from WPs 1–3 into a unified ecosystem. It orchestrates experimental validation and stress-testing on large-scale distributed research infrastructures, such as SLICES.