Security-aware dynamic clustering of distributed network nodes considering their cryptographic state

Abstract

This paper addresses the problem of security-aware dynamic clustering of nodes in distributed networks, taking into account their cryptographic state. The need to consider not only the resource and network characteristics of nodes but also the compliance of their cryptographic configurations with an established policy is substantiated, particularly during the updating of cryptographic mechanisms. The aim of this study is to develop and experimentally evaluate a clustering method designed to reduce the local concentration of nodes that do not comply with current cryptographic requirements. A cryptographic state model is proposed that classifies nodes into three categories: those that comply with the established policy, those in a transitional state, and those that do not meet its requirements. A normalized cluster formation criterion is developed that combines resource characteristics with the cryptographic states of nodes. To adapt the cluster structure, monitoring and reconfiguration are performed when a specified threshold for the concentration of non-compliant nodes is exceeded. Simulation experiments were conducted to compare the proposed method with baseline clustering under different distributions of cryptographic states. Across four experimental scenarios, the proposed method reduced the mean maximum concentration of cryptographically non-compliant nodes by between forty-three point seventy-eight and fifty-eight point fifty-seven percent. A scalability study involving networks ranging from fifty to one thousand nodes demonstrated a reduction in this metric of between thirty-eight point seventy and fifty-seven point forty-two percent. Statistical testing confirmed that the observed differences were significant. Dynamic simulations demonstrated the ability to adapt the cluster structure to changes in cryptographic states while revealing substantial overhead associated with global reconfiguration. Future research will focus on developing local reconfiguration mechanisms and reducing the number of node relocations.