EdgeAIMetric Benchmark for Performance and Energy Consumption Evaluation of Single-Board Computers for AI Edge Environments

• Gustavo Martins Conceição UNISINOS
• Rodrigo da Rosa Righi UNISINOS

DOI: https://doi.org/10.5753/eradrs.2025.6767

Abstract

With the exponential growth of IoT devices, large amounts of data are generated, causing latency and security issues. Edge computing mitigates these challenges by enabling real-time AI execution closer to data sources. Single Board Computers (SBCs) play a key role in this paradigm, but their performance and energy consumption require thorough evaluation. This paper presents EdgeAIMetric, a benchmark designed to assess SBCs running AI algorithms. It measures CPU, RAM, and energy consumption while executing Decision Tree, K-Means, Naïve Bayes, SVM, and CNN on different datasets. Experimental results highlight trade-offs between performance and efficiency, guiding AI deployment at the edge.

Keywords: Evaluation, Measurement, Performance Prediction

References

Baller, S. P., Jindal, A., Chadha, M., and Gerndt, M. (2021). Deepedgebench: Benchmarking deep neural networks on edge devices. In 2021 IEEE International Conference on Cloud Engineering (IC2E), pages 20–30.

Grambow, M., Lehmann, F., and Bermbach, D. (2019). Continuous benchmarking: Using system benchmarking in build pipelines. In 2019 IEEE International Conference on Cloud Engineering (IC2E), pages 241–246.

Hadidi, R., Cao, J., Xie, Y., Asgari, B., Krishna, T., and Kim, H. (2019). Characterizing the deployment of deep neural networks on commercial edge devices. In 2019 IEEE International Symposium on Workload Characterization (IISWC), pages 35–48.

Muhoza, A. C., Bergeret, E., Brdys, C., and Gary, F. (2023). Power consumption reduction for IoT devices thanks to edge-AI: Application to human activity recognition. Microprocessors and Microsystems, volume 24, page 100930.

Sanchez Sánchez, P. M., Jorquera Valero, J. M., Huertas Celdrán, A., Bovet, G., Gil Pérez, M., and Martínez Perez, G. (2023). Lwhbench: A low-level hardware component benchmark and dataset for single board computers. Microprocessors and Microsystems, volume 22, page 100764.

Tamburello, M., Caruso, G., Adami, D., and Giordano, S. (2023). Experimental comparison between SBC and FPGA for embedded neural network acceleration. In ICC 2023 - IEEE International Conference on Communications, pages 6078–6083.