A Linear Probabilistic Resilience Model for Securing Critical Infrastructure in Industry 5.0

Abstract

Critical infrastructures are designed for securing interconnecting networks from different influencing factors such as adversaries, unauthorized platoons, cyber threats, etc. These infrastructure hosts include human, physical elements, and cyber paradigms. The vital part is cyber resilience against weak and volatile authentication and security administrations. For strengthening cyber security, this article introduces the Artificial Intelligence-induced Constructive Resilience Model (AI-CRM). The proposed model accounts for the security requirements of the adversary impacting infrastructure elements based on probability. This probability is computed using previous adversary impacts on infrastructure failures and session drops in handling operational services. The computation for linearity or stagnancy is validated using a recurrent learning paradigm over different service transitions. The resilience is improved by augmenting security measures that are identified as an output of linear impacts over the services. Based on the linear incremental probability the resilience between two successive service transitions is computed. Identifying the non-linear or stagnant probability is the converging solution of recurrent learning. The recurrent learning optimizes the stagnancy and linear impact (probability) by repeatedly computing the failures and drops due to adversary injection. This improves resilience through security augmentations and modifications. This model is analyzed using adversary detection ratio, session drops, infrastructure failures, time lag, and service dissemination ratio.

Keywords

• Critical infrastructure
• linear processing
• recurrent learning
• resilience