Effective Machine Learning-Based Spam Detection for Internet of Things Gadgets

Abstract

In the realm of IoT, a growing challenge is the proliferation of spam and malicious communication targeting IoT (Internet of Things) devices. IoT spam refers to disruptive, irrelevant, or harmful data traffic that can overwhelm communication channels, leading to resource depletion, network congestion, and potential security vulnerabilities. The impact of IoT spam ranges from disrupting device communication to compromising the security and privacy of sensitive data. Attackers often exploit vulnerabilities in sophisticated IoT-based systems to perpetrate these threats. This article proposes enhancing IoT device security by detecting spam using Machine Learning (ML) techniques. A Spam Detection System for IoT (Internet of Things)using an artificial intelligence framework is proposed to achieve this objective. This framework evaluates three ML (Machine Learning) models employing diverse algorithms and a wide range of input feature sets. Each model is evaluated based on its accuracy score derived from processed data features, which gauges the resilience of IoT devices under various conditions. Specifically, the proposed framework integrates a Hybrid Xgboost and BGLM model, leveraging their complementary strengths to improve detection efficacy. By implementing this approach, IoT (Internet of Things) systems can better defend against spam attacks, ensuring more robust and secure operations in diverse IoT environments.