Design and simulation of material-integrated distributed sensor processing with a code-based agent platform and...

Area of Science:

• Computer Science
• Distributed Systems
• Artificial Intelligence

Background:

• Multi-agent systems (MAS) offer decentralized data processing for resource-constrained environments like sensor networks.
• Existing approaches face challenges in scalability and adaptability for complex tasks.

Purpose of the Study:

• To introduce a novel agent-based approach for decentralized data processing in distributed systems.
• To enable efficient information extraction and material integration in structural monitoring applications.

Main Methods:

• Development of agent behavior using a dynamic activity-transition graph (ATG) model.
• Implementation of self-modifying and migrating agent program code (containers).
• Utilization of a microchip-level, multi-core stack machine for agent execution, simulating token-queue processing.

Main Results:

• Achieved small-sized agent program code, low system complexity, and high system performance.
• Demonstrated agent code modification via code morphing and network migration.
• Validated the approach through architectural-level simulation of agent platforms in a large-scale distributed sensor network.

Conclusions:

• The proposed MAS approach provides a scalable and adaptable solution for distributed data processing.
• Novel self-contained, self-modifying agent code and microchip platforms enhance efficiency and performance.
• The token-queue-based agent processing is compatible with heterogeneous networks and suitable for structural monitoring.