Orchestrating Heterogeneous Devices and AI Services as Virtual Sensors for Secure Cloud-Based IoT Applications.

Area of Science:

• Computer Science
• Artificial Intelligence
• Internet of Things

Background:

• The proliferation of cloud, edge, and IoT resources necessitates integrated approaches for data management and analysis.
• Combining symbolic Artificial Intelligence (AI) with machine learning (ML) offers new possibilities for intelligent systems.

Purpose of the Study:

• To present a novel framework for semantic orchestration of Internet of Things (IoT) use cases.
• To demonstrate the utility of virtual sensors, powered by ML services, for abstracting IoT data.
• To address challenges of missing or scarce training data in ML models for IoT applications.

Main Methods:

• Development of a thing registry and an agent-based orchestration framework.
• Implementation of virtual sensors using ML services for data abstraction.
• Application of ML models for activity recognition.
• Development of methods to handle limited training data.

Main Results:

• Successful semantic orchestration of IoT use cases across federated cloud environments.
• Demonstration of virtual sensors effectively mediating between instance and semantic levels.
• Validation of ML models for activity recognition in an assisted living scenario.
• An effective approach for training ML models with scarce data was presented.

Conclusions:

• The proposed framework enables semantic orchestration of IoT, enhancing data utilization across distributed environments.
• Virtual sensors provide a valuable abstraction layer for integrating ML into IoT systems.
• The approach is effective in real-world scenarios, such as assisted living, even with data limitations.