COPRAS-Based Leader Election for Distributed Computing: A Multi-Criteria Decision Making Approach

Abstract Leader election has long been viewed as one of the core challenges in distributed systems, mainly because the participating nodes must eventually agree on a single coordinator that oversees tasks such as synchronization, resource sharing, and failure handling. Earlier approaches---for instance, the Bully algorithm or the classical ring-based techniques---work reasonably well in stable and predictable networks, but they depend heavily on fixed identifiers or predefined priorities. As a result, they do not adapt well to the heterogeneous and constantly shifting conditions seen in present-day cloud, edge, and IoT deployments. In this work, we explore a leader election approach that builds on the Complex Proportional Assessment (COPRAS) method from multi-criteria decision-making. The idea is straightforward: each potential leader is evaluated using a mix of beneficial attributes (such as CPU strength, remaining energy, or reliability) and non-beneficial attributes (including latency or communication cost). Depending on the scenario, the weighting of these criteria may come from expert knowledge or an entropy-driven estimation process. These evaluations are then combined with a lightweight message-passing routine that still operates correctly even when the system is asynchronous or experiences partial failures. Through formal reasoning, we show that the proposed method satisfies the basic properties expected from a leader election protocol, including termination, uniqueness, agreement, and validity. The computational cost per node grows on the order of \((O(mn))\), while the communication cost in the worst case reaches \((O(n^{2} m))\) for \((n)\) nodes and \((m)\) evaluation criteria. To observe how this behaves in practice, we carried out MPI-based simulations using a Google Colab setup. Across several network structures, the COPRAS-based strategy converged roughly \((15%)\) faster and achieved around \((20%)\) higher utility scores for the chosen leader compared to both TOPSIS-driven selection and simpler identifier-based methods. Overall, the study indicates that COPRAS offers a practical, interpretable, and scalable route for leader election in diverse distributed environments.