mIoT: Metamorphic IoT Platform for On-Demand Hardware Replacement in Large-Scaled IoT Applications

Area of Science:

• Computer Science
• Electrical Engineering
• Embedded Systems

Background:

• The Internet of Things (IoT) paradigm is shifting towards edge computing due to server bottlenecks, security concerns, and increasing data volumes.
• Edge devices require enhanced computational capabilities for complex applications like AI and signal processing, but face hardware limitations.
• Current edge solutions lack flexibility due to fixed hardware functions, hindering performance and operational efficiency.

Purpose of the Study:

• To propose a novel metamorphic IoT (mIoT) platform enabling dynamic hardware acceleration on edge devices with limited resources.
• To address the need for flexible processing and computation in diverse IoT applications through on-demand hardware reconfiguration.
• To reduce data transmission to servers by reconfiguring hardware directly at the edge.

Main Methods:

• Developed a metamorphic IoT (mIoT) platform featuring reconfigurable edge devices based on RISC-V architecture.
• Implemented a server-based management system that utilizes a probabilistic value, 'callability,' to control edge device hardware reconfiguration.
• Proposed an on-demand transmission and reconfiguration strategy for required hardware at the edge, minimizing overhead.

Main Results:

• The callability-based mIoT platform successfully provided required hardware to edge devices in real time.
• Demonstrated the capability of performing diverse functions using limited hardware resources on edge devices.
• Achieved significant reduction in power consumption, a critical constraint for IoT devices.

Conclusions:

• The mIoT platform offers a flexible and efficient solution for edge computing in the Internet of Things.
• Dynamic hardware reconfiguration based on callability enhances the processing capabilities of resource-constrained edge devices.
• The proposed approach effectively balances computational demands with power efficiency in IoT ecosystems.