Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{193837,
        author = {Emmanuel Usen and Emmanuel Usen},
        title = {Review of the Adoption of an Energy-Efficient Security Mechanism for Neural Networks},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {10},
        pages = {2305-2323},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=193837},
        abstract = {Neural networks are increasingly embedded in intelligent and connected vehicles to support perception, decision-making, and cooperative functions. While their security is critical for safety, vehicular deployments operate under strict energy and resource constraints. Existing studies broadly address neural network security and energy efficiency in isolation, limiting their practical relevance for vehicle systems. This paper examines how security mechanisms for neural networks can be designed and evaluated with explicit consideration of energy constraints in intelligent and connected vehicles. A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-aligned structured literature review was conducted across major scientific databases, yielding 47 relevant studies. The selected works were analysed using thematic synthesis and classified according to the neural network lifecycle, encompassing model-level, data- and training-level, and deployment-level security mechanisms. Energy implications were assessed based on reported metrics or inferred computational and communication overheads. The review shows that many effective neural network security mechanisms impose substantial energy costs that challenge on-board and edge-based vehicular deployment. Model-level approaches offer energy savings but require robustness-aware design; training-level mechanisms remain computationally intensive; and deployment-level strategies provide practical trade-offs through secure, communication-efficient inference. No single class of mechanisms sufficiently balances security and energy efficiency on its own. The paper then proposes a conceptual framework that integrates security objectives, energy constraints, and vehicular deployment contexts. The framework guides the development of secure, energy-aware neural networks suitable for intelligent and connected vehicle systems by treating energy consumption as a first-order design constraint.},
        keywords = {Adversarial robustness; Edge-based vehicular computing; Energy-efficient AI; Intelligent and connected vehicles; Neural network security.},
        month = {March},
        }