Connecting the Dots: A Systematic Literature Review of Explainable AI, Cybersecurity, Human-Centered Design and Edge Computing

AI adoption in cybersecurity has accelerated sharply, driven by generative AI and large language models. These technologies promise to transform threat detection — but they introduce profound challenges around explainability, trust, and deployment feasibility in resource-constrained environments. The evidence addressing those challenges was scattered across four research communities that rarely cite each other: explainable AI, cybersecurity operations, human-centered design, and edge computing.

Researchers and practitioners designing AI-based security systems — especially those who must make them explainable to human operators and deployable outside the data center, where compute, memory, and connectivity are constrained.

First author, working with co-authors Fabrizio Benelli, Mario Caronna, Giulia Palma, and Antonio Rizzo at the University of Siena. The review was designed as a hybrid: a qualitative systematic literature review combined with a mapping study, so that it both synthesizes what the empirical evidence says and charts where it is thin.

• Strict adherence to the PRISMA 2020 guidelines for systematic reviews — protocol, screening, and reporting.
• Scoping at the nexus: rather than reviewing each field separately, the study filters for empirical evidence at the precise intersection of XAI methods, cybersecurity operations, human-centered design, and edge-computing constraints.
• Hybrid design: systematic synthesis of the included studies plus a mapping of the research landscape across the four domains.

• Reviewing the intersection, not the union: the deliberate choice to study where all four domains overlap is what makes the review useful to system designers, who face all four constraints at once.
• PRISMA 2020 as a hard constraint: full methodological transparency so the review itself is auditable — the same principle my applied work asks of AI systems.

Published in the Journal of Cybersecurity and Privacy (MDPI) in 2026 as my first first-author, peer-reviewed publication: a synthesis and map of the empirical evidence connecting explainable AI, cybersecurity, human-centered design, and edge computing.

• How to run a review that others can trust: PRISMA discipline — explicit protocols, documented screening decisions — is what separates a citable synthesis from an opinion piece.
• The map guides the building: this review is the theoretical backbone of my applied projects — the same four domains it connects are the ones Cracker Breaker and SAAM put into production.