Frugal Self-Optimization Mechanisms for Edge-Cloud Continuum

Area of Science:

• Computer Science
• Distributed Systems
• Artificial Intelligence

Background:

• The Edge-Cloud Continuum (ECC) faces increasing complexity due to IoT expansion and data-intensive applications.
• Current systems require innovative methods for automated and efficient management, focusing on self-* capabilities.
• Anomaly detection and adaptive sampling are crucial for optimizing data transmission and system reliability.

Purpose of the Study:

• To integrate anomaly detection and adaptive sampling into a single, lightweight, and extendable self-optimization module.
• To design a module suitable for resource-constrained devices in distributed ECC environments.
• To enhance system autonomy, operational proactiveness, and resilience of critical processes.

Main Methods:

• Implementation of density-based anomaly detection in real-time resource utilization data streams.
• Integration of a dynamic adaptive sampling technique using Probabilistic Exponential Weighted Moving Average.
• Validation using synthetic and real-world datasets, including target infrastructure samples.

Main Results:

• Demonstrated effectiveness of implemented techniques in diverse, near real-life scenarios.
• Performance comparison with state-of-the-art algorithms, highlighting advantages and limitations.
• Showcased the module's capability for resource-aware autonomic optimization.

Conclusions:

• The proposed module offers a novel perspective on resource-aware autonomic optimization for next-generation ECC.
• The module is well-suited for distributed systems with highly resource-constrained devices.
• Emphasis on frugality and real-time operation enhances the resilience of critical processes.