A Predictive Neural Network Architecture for Early Detection of Low-Rate Cyberattacks (opens in new tab)

Low-Rate Denial of Service (LDoS) attacks pose a significant challenge to IoT networks due to their subtle and prolonged nature, often evading traditional intrusion detection systems. This paper presents IDQS (Intrusion Detection via QoS Prediction), a lightweight and proactive framework for early LDoS attack detection. IDQS integrates two new key components: (i) RTP-QoS, a Recurrent Trend Predictive Neural Network that learns and forecasts futu...