Riemannian Flow Analysis for Secure Software Dependency Resolution in Microservices Architectures

Abstract

Microservice architecture bases many of its operations on the dynamic software dependency resolution, which leads to severe security threats to systems, including dependency confusion, supply-chain attacks, and malicious package injections. The conventional dependency management and anomaly detection tools are mostly based on static graph or rule-based models that are perceived to be incapable of defining the non-linear and dynamic nature of dependency interactions. The paper introduces a new framework that relies on the Riemannian flow analysis to represent the microservice dependency graphs as smooth manifolds to use the continuous flows to describe the normal resolution dynamics. The framework identifies deviations in benign dependency behavior geometric structure with lower false positives and greater precision than current methods by learning this geometric structure. The framework combines manifold embeddings, flow-based anomaly detection, and runtime policy enforcement into the containerized settings including Kubernetes. The study provides not only theoretical understanding of how Riemannian geometry can be applied in cybersecurity, but also opens a practical avenue towards the attainment of secure and scalable dependency resolution of microservices architectures.