Logo PTI Logo FedCSIS

Proceedings of the 18th Conference on Computer Science and Intelligence Systems

Annals of Computer Science and Information Systems, Volume 35

IoT and Edge Computing using virtualized low-resource integer Machine Learning with support for CNN, ANN, and Decision Trees

DOI: http://dx.doi.org/10.15439/2023F7745

Citation: Proceedings of the 18th Conference on Computer Science and Intelligence Systems, M. Ganzha, L. Maciaszek, M. Paprzycki, D. Ślęzak (eds). ACSIS, Vol. 35, pages 367376 ()

Full text

Abstract. Tiny Machine Learning is a new approach that is being used for data-driven prediction classification and regression on microcontrollers using local sensor data. The models are typically learned off-line and sent to the microcontroller for use as binary objects or frozen and converted static data. This approach is not universal or flexible. The REXA VM introduced in this work, which can virtualize embedded systems and sensor nodes and includes a general machine learning framework that supports arbitrary dynamic ANN and decision tree (DT) models, is introduced in this study. The models are delivered as text files with highly compressed program code that are enclosed in code frames with embedded data (model parameters). The VM offers fundamental computations for ANN and DT models (Microservices). Using a decompiler, models can be updated (retrained) and sent to other nodes (mobile models). It can be demonstrated that virtualization using a bytecode machine and just-in-time compiler is still appropriate and effective for extremely low-resource processors.


  1. Guo, S., Zhou, Q. , Machine Learning on Commodity Tiny Devices, Taylor & Francis, 2023
  2. Ray, P. P., A review on TinyML: State-of-the-art and prospects, Journal of King Saud University-Computer and Information Sciences, 2021, pp.1595-1623, https://doi.org/10.1016/j.jksuci.2021.11.019
  3. Wang, X., Magno, M. , Cavigelli, L., Benini, L., FANN-on-MCU: An Open-Source Toolkit for Energy-Efficient Neural Network Inference at the Edge of the Internet of Thing, https://arxiv.org/abs/1911.03314v3, 2022
  4. Banner, R. , Hubara, I., Hoffer, E., Soudry, D., Scalable Methods for 8-bit Training of Neural Networks, https://arxiv.org/abs/1805.11046, 2018
  5. Alajlan, N. N., Ibrahim, D. M., TinyML: Enabling of Inference Deep Learning Models on Ultra-Low-Power IoT Edge Devices for AI Applications, micromechanics, vol. 13, no. 851, 2022, https://doi.org/10.3390/mi13060851
  6. Jain, V., Giraldo, S., Roose, J. D., Linyan, Mei, B. B., Verhelst, M. , TinyVers: A Tiny Versatile System-on-chip with State-Retentive eMRAM for ML Inference at the Extreme Edge, https://arxiv.org/abs/2301.03537, 2023,
  7. Heiser, G., The role of virtualization in embedded systems, In Proceedings of the 1st workshop on Isolation and integration in embedded systems, 11-16 April, 2008, https://doi.org/10.48550/arXiv.2301.03537
  8. Zhang, L., Implementation of fixed-point neuron models with threshold, ramp and sigmoid activation functions, In IOP Conference Series: Materials Science and Engineering (Vol. 224, No. 1, p. 012054). IOP Publishing, 2017
  9. Bosse, S., Bornemann, S., Lüssum, B., Virtualization of Tiny Embedded Systems with a robust real-time capable and extensible Stack Virtual Machine REXAVM supporting Material-integrated Intelligent Systems and Tiny Machine Learning, https://arxiv.org/abs/2302.09002 [cs.OS], 2023, https://doi.org/10.48550/arXiv.2302.09002
  10. Bauer, M., IoT Virtualization with ML-based Information Extraction, in IEEE 7th World Forum on Internet of Things 2021, https://doi.org/10.1109/WF-IoT51360.2021.9595119
  11. Hayes, J. R. Fraeman, M. E., Williams, R. L. Zaremba, T., An architecture for the direct execution of the Forth programming language, ACM SIGARCH Computer Architecture News, 15(5), 1987, pp. 42-49. https://doi.org/10.1145/36177.36182
  12. Ghaffari, A.,. Tahaei, M. S., Tayaranian, M., Asgharian, Vahid, M., Nia, P., Is Integer Arithmetic Enough for Deep Learning Training?, Advances in Neural Information Processing Systems 35. 2022: 27402-27413.
  13. Bosse, S., Polle, C., Fast Temperature-Compensated Method for Damage Detection and Structural Health Monitoring with Guided Ultrasonic Waves and Embedded Systems, Eng. Proc. 2021, 10(1), 78; https://doi.org/10.3390/ecsa-8-11283
  14. https://github.com/bsLab/rexavm, REXA VM repository, on-line, accessed 31.7.2023