AUTOMATED COMPLEX "ELECTRONIC SIGNATURE OF THE VOTER (EPO)" AS AN ELEMENT OF THE CRITICAL INFRASTRUCTURE OF THE ELECTION PROCESS

Abstract

The article studies the automated complex "Electronic Voter Signature" (EPO) as an element of the critical information infrastructure of the election process. It is shown that modern electoral systems can no longer rely solely on paper procedures and traditional identification mechanisms, as the intensity of cyber threats aimed at undermining trust in voting results is increasing. A system model of the EPO complex is proposed, in which key modules are allocated: biometric verification of voters, cryptographic protection and signature, management of voting sessions, logging and auditing, integration with the electronic voting system, as well as security monitoring subsystems based on Zero Trust, IDS (intrusion detection systems) and SIEM (centralized event correlation) approaches. To assess the reliability and stability of the complex, probabilistic models of reliability theory, indicators of average uptime (MTBF) and mean recovery time (MTTR), availability factor and generalized stability function were used. Special attention is paid to modeling cascade failures in the network of district nodes interacting through secure channels, using graph models and simulation scenarios. To increase the effectiveness of cyber defense, the use of machine learning methods, in particular, deep architectures CNN+LSTM, AE+LSTM and Byte2Image transformations for analyzing network traffic and event logs is proposed. It has been demonstrated that the combination of redundant architectural solutions, the Zero Trust concept and intelligent anomaly detection systems allows achieving a significantly higher level of availability and resilience of the EPO complex compared to the basic configurations. A method for protecting the biometric identification and authentication unit by human face from spoofing attacks is proposed using algorithms for converting color image spaces into YCrCb and CIE L*u*v* and analyzing attack signs using the ETC classifier. The normative aspects of the regulation of electoral information systems in Ukraine were discussed, the need for formal recognition of electoral systems as a component of the national critical infrastructure was emphasized. Recommendations were formulated for the integration of cyber resilience requirements into national electoral and cybersecurity regulations, as well as directions for further research on risk modeling and evaluation of the effectiveness of hybrid ML/AI solutions in the electoral process.