A study on the design methodology of TAC³ for edge computing

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Massive data, complex applications, and billions of devices challenge traditional ICT infrastructure.
• The Internet of Everything (IoE) necessitates advanced solutions for computing, network, and storage capabilities.
• Existing ICT frameworks struggle with the scale and security demands of modern digital environments.

Purpose of the Study:

• To propose TAC3 (Tile-Architecture Cluster Computing Core) as a lightweight, accelerated Edge Computing Node (ECN).
• To address the challenges posed by complex application requirements and Zettabyte-scale data in the IoE.
• To enhance the capabilities and security of computing, network, and storage in the face of massive connectivity.

Main Methods:

• Utilized a "description-synthesis-simulation-optimization" design methodology.
• Employed executable description specification and polymorphous parallelism for Design Space Exploration (DSE) of the Tile-Architecture ECN.
• Configured edge computing environments and optimized typical application scenarios like Convolutional Neural Networks (CNNs) and image processing.

Main Results:

• TAC3 was designed and simulated as a lightweight accelerated ECN.
• The proposed methods effectively addressed challenges in network bandwidth, end-cloud delay, and privacy security for IoE data.
• Optimized edge computing environments demonstrated improved performance for AI-driven applications.

Conclusions:

• TAC3 provides a viable solution for the computational and security demands of the IoE.
• The "Edge-Cloud complements each other, and Edge-AI energizes each other" philosophy is presented as a future IoE principle.
• Edge computing, exemplified by TAC3, is crucial for managing massive data and ensuring privacy in interconnected systems.