Intelligent Lightweight Real-Time DOS/DDOS Attack Detection and IOT-Based Alerting Framework with Performance Evaluation for Small-Scale Network Environments
Prof. Shital S. Patil∗
Department of Information Technology,
SVIT Nashik, Maharashtra, India
Mr. Om Prashant Raut†
Department of Information Technology,
SVIT Nashik, Maharashtra, India
Mr. Karan Kishor Targe‡,
Department of Information Technology,
SVIT Nashik, Maharashtra, India
Ms. Laxmi Punamchand Kasar§
Department of Information Technology,
SVIT Nashik, Maharashtra, India
Ms. Sakshi Anil Raut¶
Department of Information Technology,
SVIT Nashik, Maharashtra, India
Email: {rautom405@gmail.com, karantarge5@gmail.com ,laxmikasar7@gmail.com ,sakshiraut390@gmail.com,}
ABSTRACT
The proliferation of Internet of Things (IoT) devices in small-scale environments—such as smart homes, small offices, and clinic setups—has introduced significant security vulnerabilities, particularly to Denial of Service (DoS) and Distributed Denial of Service (DDoS) attacks. Traditional intrusion detection systems remain impractical for these settings due to their high computational overhead, cost, and complexity. This paper presents a comprehensive framework for intelligent, lightweight, real-time DoS/DDoS attack detection coupled with an IoT-based alerting mechanism, specifically designed for resource-constrained, small-scale network environments. The framework synthesizes recent advances in lightweight machine learning architectures—including Modified Gated Recurrent Units (MGRU), hybrid LSTM-CNN models, and TinyML-optimized classifiers—with distributed collaborative intelligence for threat validation. We evaluate the framework's performance across multiple dimensions: detection accuracy (96-100%), response time (1-125 ms), memory footprint (82KB-2.05GB depending on node type), and computational efficiency. The proposed architecture achieves 99%+ detection accuracy for known attack vectors while operating within the strict resource limits of ESP32-class devices and Raspberry Pi–based edge coordinators. Additionally, we present a tiered alerting mechanism that leverages low-cost IoT components (MQTT brokers, OLED displays, LED indicators) to provide real-time network status visualization. Performance evaluation demonstrates that the framework reduces CPU utilization by 77% and memory consumption by 92% compared to centralized alternatives, with implementation costs under €200—making enterprise-grade security accessible to small-scale deployments.
Keywords: DDoS Detection, IoT Security, Lightweight Machine Learning, TinyML, Real-Time Alerting, Small-Scale Networks, Edge Computing, Intrusion Detection System
