Experimental evaluation of an adaptive hybrid architecture for Blockchain based digital identity management

Abstract

The management of digital identities unfolds in a context of persistent tension between security, performance, and energy consumption, particularly when centralized architectures are compared with blockchain-based approaches. In this setting, the present study introduces and evaluates an adaptive hybrid architectural model that integrates centralized and decentralized processing and switches dynamically according to system load, risk level, and operational criticality. The research is structured as an applied, quantitative study with a comparative experimental design, in which three functionally equivalent environments, namely centralized, decentralized, and hybrid, were implemented, and tests were conducted using 500 digital identities under five concurrency levels (50, 100, 200, 500, and 1000 requests). Latency, CPU utilization, energy consumption, and a computational energy efficiency metric (EPR), expressed in Wh per transaction, were systematically recorded. The results show that the hybrid model maintains global latency and EPR values that are very close to those of the centralized architecture, whereas the decentralized architecture generally exhibits higher energy consumption per transaction and greater CPU demand, even though its EPR improves under extreme load scenarios. Overall, the findings support the view that adaptive hybrid architectures can balance traceability and efficiency and emerge as a viable alternative for digital identity systems operating under variable loads and computational sustainability constraints.