Swarm-Intelligence–Guided Clustering for Distributed Data Repositories with Communication-Aware Constraints

• Rendra Saputra

  Universitas Bosowa, Jalan Urip Sumoharjo No. 242, Departemen Sistem Informasi, Makassar, Sulawesi Selatan, Indonesia

  Author

• Yudha Firmansyah

  Universitas Kadiri, Jalan Selomangleng No. 1, Departemen Teknik Komputer, Kediri, Jawa Timur, Indonesia

  Author

Distributed data repositories arise in settings where data are naturally partitioned across administrative domains, geographic regions, or edge devices, and where moving raw records is undesirable due to cost, latency, or policy. Clustering is frequently used in such environments to build shared structure for indexing, summarization, anomaly triage, and downstream learning, yet standard clustering formulations typically assume a centralized view of the dataset and ignore the mechanisms by which cluster state must be coordinated. This paper studies clustering over distributed repositories under communication-aware constraints, focusing on regimes where bandwidth budgets, heterogeneous links, and asynchronous participation materially affect achievable clustering quality. The core contribution is a swarm-intelligence–guided approach that treats cluster parameters as a population of candidate solutions that evolve through locally evaluated fitness signals, while explicitly accounting for the communication required to maintain and refine those solutions. Communication-aware constraints are incorporated through a constrained objective with explicit per-link and per-round budgets, delay-sensitive penalties, and message-size–dependent accounting of overhead. The resulting method uses distributed swarm updates, repository-level sub-swarms, and intermittent consensus to trade off distortion reduction against coordination cost. Analytical considerations are provided for stability and constraint satisfaction in stochastic and asynchronous settings, along with complexity characterizations that separate computation from network overhead. An evaluation framework is described to quantify clustering quality, constraint adherence, and communication efficiency across synthetic and repository-derived workloads.

• PDF

Copyright (c) 2026 International Journal of Science, Engineering and Applied Technologies

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.