Clustering Algorithms on Low-Power and High-Performance Devices for Edge Computing Environments.

Area of Science:

• Artificial Intelligence
• Edge Computing
• Embedded Systems

Background:

• The integration of Artificial Intelligence (AI) with Edge Computing enables decentralized decision-making at the network periphery.
• There's a growing trend of developing integrated sensor and computing devices for in-situ data processing.
• These edge devices prioritize low energy consumption over high computational power.

Purpose of the Study:

• To investigate the implementation of computationally intensive clustering algorithms on parallel, low-energy edge devices.
• To evaluate the performance and energy efficiency trade-offs of different edge computing hardware for AI tasks.

Main Methods:

• Implementation of clustering algorithms on two distinct edge devices: UDOO X86 Advanced+ (quad-core) and NVIDIA Jetson Nano (GPU-based).
• Comparative analysis of performance metrics and energy consumption for each device.
• Evaluation of the suitability of these devices for edge AI applications.

Main Results:

• Both UDOO X86 Advanced+ and NVIDIA Jetson Nano demonstrated viable performance for clustering tasks.
• The evaluated edge devices achieved a more favorable performance-to-energy consumption ratio compared to traditional high-end computing solutions.
• The study highlights the potential of specialized edge hardware for efficient AI deployment.

Conclusions:

• Edge computing platforms, particularly those with parallel processing capabilities like the NVIDIA Jetson Nano, are effective for deploying AI clustering algorithms.
• These low-energy devices offer a practical and efficient solution for real-time data analysis in resource-constrained environments.
• The findings support the adoption of edge AI for applications requiring decentralized intelligence and power efficiency.