<![CDATA[Validators in Federated Byzantine Agreement (FBA) systems, such as the Stellar network, are essential to guaranteeing transaction security and network liveness. However, the metrics currently used to determine validator trust are based on aggregated data (active rate, agreement rate, latency) that may not reflect the detailed behavioral changes required for accurate reputation assessment. This paper analyzes a dataset of 7,150 observations from 55 Stellar validators spanning 133 days and identifies five fundamental limitations of aggregated metrics: parameter poverty, zero variance in agreement rate, minimal variance in active rate, temporal insensitivity, and live data sparsity. Based on these findings, this paper proposes a framework for extracting 12 behavioral parameters from SCP.log files including ballot consistency, nomination success rates, timeout frequency, and equivocation attempts for enhanced EigenTrust-based reputation assessment. While full implementation requires continuous SCP.log collection, the proposed framework addresses the identified limitations by specifying the behavioral parameters necessary for dynamic validator trust assessment in FBA systems.]]>
Copyrights © 2026
