Distributed computing system based on mobile devices

This paper examines the architecture of a distributed computing system built on heterogeneous mobile devices and employing a combined dynamic load balancing method. This approach is focused on wireless environments where the composition of nodes and their performance vary over time. The performance of smartphones as computing nodes is analyzed, and the factors limiting their effectiveness are investigated: heterogeneity of hardware platforms, thermal throttling, heterogeneity of computing cores, and background activity load. An algorithm is proposed that combines a static assessment of node capacity and dynamic adjustment of performance factors taking into account the frequency, temperature, and current processor load. The algorithm incorporates a fault-tolerant subtask redistribution mechanism: if a node is disconnected or freezes, unfinished subtasks are automatically returned to the queue and assigned to other workers. The proposed approach ensures adaptation of load distribution to the current state of computing nodes, maintaining stability of overall performance during fluctuations in their resources. Experimental testing was performed on a set of smartphones of different classes, using a task without inter-node data exchange as the test load. The experimental evaluation confirms that the developed method significantly reduces task execution time and minimizes load variance compared to static approaches.

1. Rodríguez J.M., Mateos C., Zunino A. Are smartphones really useful for scientific computing? In: Advances in New Technologies, Interactive Interfaces and Communicability: Second International Conference, ADNTIIC 2011, 05–07 December 2011, Huerta Grande, Argentina. Berlin, Heidelberg: Springer; 2012. P. 38–47. https://doi.org/10.1007/978-3-642-34010-9_4

2. Büsching F., Schildt S., Wolf L. DroidCluster: Towards smartphone cluster computing – The streets are paved with potential computer clusters. In: 2012 32nd International Conference on Distributed Computing Systems Workshops, 18–21 June 2012, Macau, China. IEEE; 2012. P. 114–117. https://doi.org/10.1109/ICDCSW.2012.59

3. Arslan M.Y., Singh I., Singh Sh., et al. Computing while charging: building a distributed computing infrastructure using smartphones. In: CoNEXT '12: Proceedings of the 8th International Conference on Emerging Networking Experiments and Technologies, 10–13 December 2012, Nice, France. New York: ACM; 2012. P. 193–204. https://doi.org/10.1145/2413176.2413199

4. Shiraz M., Gani A., Khokhar R.H., Buyya R. A review on distributed application processing frameworks in smart mobile devices for mobile cloud computing. IEEE Communications Surveys & Tutorials. 2013;15(3):1294–1313. https://doi.org/10.1109/SURV.2012.111412.00045

5. Balabaev S.A., Lupin S.A., Telegin P.N., Shabanov B.M. Increasing the PC computing power: integration with a distributed smartphone system. Software & Systems. 2024;(4):504–513. (In Russ.). https://doi.org/10.15827/0236-235X.148.504-513

6. Isaev F.I., Isaeva G.N. Analysis of limited mobile networks and the potential of distributed mobile computing. News of the Kabardino-Balkarian Scientific Center of RAS. 2025;27(4):24–34. (In Russ.). https://doi.org/10.35330/1991-6639-2025-27-4-24-34

7. Qin Y., Zeng G., Kurachi R., Matsubara Y., Takada H. Execution-variance-aware task allocation for energy minimization on the big.LITTLE architecture. Sustainable Computing: Informatics and Systems. 2019;22:155–166. https://doi.org/10.1016/j.suscom.2018.10.001

8. Dolgov A.A. Deploying a grid system from mobile devices on the BOINC platform. In: Cloud and distributed computing systems in electronic management: Proceedings of the 3rd International Scientific and Technical Conference, 29 November – 02 December 2022, Pereslavl-Zalessky, Russia. Pereslavl-Zalessky: Program Systems Institute of the RAS; 2023. P. 24–29. (In Russ.).

9. Krioukov A., Mohan P., Alspaugh S., et al. NapSAC: Design and implementation of a power-proportional web cluster. In: Green Networking '10: Proceedings of the first ACM SIGCOMM workshop on Green networking, 30 August 2010, New Delhi, India. New York: ACM; 2010. P. 15–22. https://doi.org/10.1145/1851290.1851294

10. Khaing M.T., Lupin S., Thu A. Evaluating the effectiveness of load balancing methods in distributed computing systems. International Journal of Open Information Technologies. 2021;9(11):30–36. (In Russ.).