A.Yu. Petrov1

1 Saint-Petersburg State University of Aerospace Instrumentation (St. Petersburg, Russia)

1 palexeq1998@yandex.ru

This paper addresses the problem of efficient computational load distribution in distributed computing systems operating under conditions of hardware heterogeneity, varying task volumes, and communication constraints. In such environments, conventional static or uniform scheduling approaches often fail to ensure balanced resource utilization and minimum execution time, particularly when individual nodes differ in performance characteristics. As a result, the development of adaptive multi-objective optimization methods capable of simultaneously accounting for throughput, node workload, and communication delays becomes an important research task.

The aim of the study is to develop and experimentally validate the software implementation of the hybrid multi-objective Adaptive Swarm-Genetic Resource Allocation Optimization Method (ASGRAOM) algorithm designed for optimizing load allocation in distributed systems, as well as to assess its effectiveness on a real hardware platform. The proposed algorithm combines the modified Multi-Objective Particle Swarm Optimization method with genetic algorithm operators within a unified adaptive optimization cycle. A distinctive feature of hybrid algorithm is the dynamic switching between the swarm-based search phase and the evolutionary diversification phase based on the analysis of Pareto front stagnation. This mechanism makes it possible to preserve convergence speed while preventing premature trapping in local compromises.

The study presents the architecture of the algorithm, the principles of maintaining an external Pareto archive, and the logic of selecting non-dominated scheduling solutions according to multiple performance criteria. The software implementation of ASGRAOM was integrated into an experimental distributed computing demonstrator built on Raspberry Pi Pico microcontrollers connected via the I2C bus in a centralized master–slave configuration. Within this platform, the optimization module executed on the host computer generated task allocation plans, while the microcontroller nodes performed parallel computational subtasks and returned the results for further analysis.

The experimental evaluation was carried out using several scenarios that differed in the number of computing nodes, the size of the computational workload, and the degree of system heterogeneity. The results demonstrated that hybrid algorithm provides a more balanced load distribution, reduces idle time of faster nodes, and improves the overall execution efficiency of the distributed system. In a number of scenarios, the proposed algorithm reduced task execution time by approximately 8% compared with a naive uniform distribution strategy while maintaining the same computational accuracy. The obtained results also confirmed the scalability of the approach and its stability under heterogeneous hardware conditions.

The practical significance of the research lies in the possibility of applying the proposed algorithm to the design and modernization of distributed computing infrastructures, embedded computing networks, and specialized hardware-software systems requiring improved performance, adaptability, and reliability of resource utilization.

Petrov A.Yu. Software implementation and analysis of experimental results of the hybrid ASGRAOM algorithm // Achievements of Modern Radioelectronics. 2026. V. 80. № 9. P. 94−106. DOI: https://doi.org/10.18127/j20700784-202609-10

1. Xu C.Z., Lau F.C.M. Load Balancing in Parallel Computers: Theory and Practice. Dordrecht: Kluwer Academic Publishers. 1997. 266 p.
2. Eager D.L., Lazowska E.D., Zahorjan J. Adaptive load sharing in homogeneous distributed systems. IEEE Transactions on Software Engineering. 1986. V. SE-12. № 5. P. 662–675.
3. Casavant T.L., Kuhl J.G. A taxonomy of scheduling in general-purpose distributed computing systems. IEEE Transactions on Software Engineering. 1988. V. 14. № 2. P. 141–154.
4. Harchol-Balter M. Performance Modeling and Design of Computer Systems. Cambridge: Cambridge University Press. 2013. 576 p.
5. Coello Coello C.A. An updated survey of GA-based multiobjective optimization techniques. ACM Computing Surveys. 2000. V. 32. № 2. P. 109–143.
6. Deb K. Multi-Objective Optimization Using Evolutionary Algorithms. Chichester: Wiley. 2001. 497 p.
7. Pis`mennaya V.A., Panteleev A.V. Programmny`j kompleks poiska optimal`nogo upravleniya nelinejny`mi determinirovanny`mi dinamicheskimi sistemami s pomoshh`yu memeticheskogo algoritma. Uspexi sovremennoj radioe`lektroniki. 2016. № 2. S. 183–186. (in Russian).
8. Zitzler E., Deb K., Thiele L. Comparison of multiobjective evolutionary algorithms: Empirical results. Evolutionary Computation. 2000. V. 8. № 2. P. 173–195.
9. Goldberg D.E. Genetic Algorithms in Search, Optimization and Machine Learning. Boston: Addison-Wesley. 1989. 412 p.
10. Katoch S., Chauhan S.S., Kumar V. A review on genetic algorithm: past, present, and future. Multimedia Tools and Applications. 2021. V. 80. P. 8091–8126.
11. Kennedy J., Eberhart R. Particle swarm optimization. Proceedings of IEEE International Conference on Neural Networks. 1995. V. 4. P. 1942–1948.
12. Gad A.G. Particle Swarm Optimization Algorithm and Its Applications: A Systematic Review. Archives of Computational Methods in Engineering. 2022. V. 29. P. 2531–2561.
13. Osika Z., Zatarain Salazar J., Roijers D.M., Oliehoek F.A., Murukannaiah P.K. What Lies beyond the Pareto Front? A Survey on Decision-Support Methods for Multi-Objective Optimization. Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI-23). 2023. P. 6741–6747
14. Chuyanov G.A., Kos`yanchuk V.V., Sel`vesyuk N.I. Perspektivy` razvitiya kompleksov bortovogo oborudovaniya na baze integrirovannoj modul`noj avioniki. Izvestiya YuFU. Texnicheskie nauki. 2013. №3 (140). S. 55–62. (in Russian).
15. Vanderleest S.H., Matthews D.C. Incremental Assurance of Multicore Integrated Modular Avionics (IMA). IEEE/AIAA 40th Digital Avionics Systems Conference (DASC). San Antonio. 2021. S. 1–9.
16. Eveleens R.L.C. Integrated Modular Avionics Development Guidance and Certification Considerations. National Aerospace Laboratory NLR. Amsterdam. 2006. 19 s.
17. Petrov A.Yu., Olenev V.L. Formalizaciya gibridnogo metoda raspredeleniya zadach s ispol`zovaniem setej Petri. Nauchno-texnicheskij vestnik Povolzh`ya. 2025. № 6. S. 228–232. (in Russian).
18. Petrov A.Yu. Gibridny`j Algoritm PSO-GA v vy`chislitel`ny`x sistemax. Ae`rokosmicheskoe priborostroenie i e`kspluatacionny`e texnologii. Mater. Shestoj Mezhdunar. nauch. konf. Sankt-Peterburg, 08–25 aprelya 2025 goda. SPb.: Sankt-Peterburgskij gosudarstvenny`j universitet ae`rokosmicheskogo priborostroeniya. 2025. S. 220–223. (in Russian).
19. Neto A.J.A., Carneiro Neto J.A., Moreno E.D. The development of a low-cost big data cluster using Apache Hadoop and Raspberry Pi. A complete guide. Computers and Electrical Engineering. 2022. V. 104. Pt. A.
20. Da Silva L.F., Lima J.V.F. An evaluation of relational and NoSQL distributed databases on a low-power cluster. The Journal of Supercomputing. 2023. V. 79. Iss. 12. P. 13402–13420.
21. Zhang Z.-D., Yu D.-Y., Ibhadode O., Meng L., Gao T., Zhu J.-H., Zhang W.-H. TopADDPi: An Affordable and Sustainable Raspberry Pi Cluster for Parallel-Computing Topology Optimization. Processes. 2025. V. 13. № 3.
22. Kiryushkin S.B. Osobennosti realizacii protokola obmena danny`mi dlya sistem upravleniya aktivnoj fazirovannoj antennoj reshetkoj. Uspexi sovremennoj radioe`lektroniki. 2016. № 2. S. 166–170. (in Russian).
23. Gridin V.N., Anisimov V.I., Vasil`ev S.A. Metod postroeniya komponentov raspredelennoj sistemy` dlya rascheta stacionarnogo rezhima e`lektronny`x sxem na osnove mikroservisov s polnodupleksny`m kanalom svyazi. Uspexi sovremennoj radioe`lektroniki. 2021. T. 75. № 8. S. 20–29. (in Russian).