Artificial Intelligence for IoT Network Security: A Systematic Analysis of Threat Detection Strategies

Abstract

Abstract– This paper presents a systematic literature review focused on threat detection strategies in Internet of Things (IoT) networks using artificial intelligence techniques, with a particular emphasis on their applicability to the Peruvian context. The PRISMA protocol was applied for the selection and analysis of studies published between 2020 and 2025, ranging from smart irrigation systems based on decision trees and random forests to advanced deep learning models (autoencoders, CNNs, LSTM) and explainable artificial intelligence (XAI) approaches for botnet detection. The review identifies the methods that achieve the highest detection rates (>95%), evaluates their scalability on resource-constrained devices, and examines the challenges arising from protocol heterogeneity and the lack of security standards. In addition, emerging solutions such as blockchain, quantum physically immutable functions (PUFs), and federated learning are described for privacy enhancement and resilience against physical and logical attacks. The findings reveal critical gaps in regional adoption, such as the lack of regulatory frameworks and limited edge computing infrastructure, which hinder the implementation of AI-based detection systems. Based on this analysis, a set of recommendations are proposed to guide the development and integration of robust solutions tailored to Peru's technological, economic, and regulatory characteristics, fostering a more resilient and reliable IoT cybersecurity architecture.