International Journal of Science, Engineering and TechnologyAuthors: Shakeeb Ahmed, Syed Zubair Yuneeb, Tejas BN, Sneha Singh, Dr.C Nandini
Abstract: – In today's hyper-connected digital landscape, organizations are confronting an escalating tide of increasingly complex and rapidly evolving cyber threats. This challenge is profoundly exacerbated across the vast and distributed systems enabled by the Internet of Things (IoT), where the sheer volume of devices and their constant communication create an expansive and often vulnerable attack surface. Traditional cybersecurity solutions, typically reliant on static, signature-based detection methods, inherently struggle to adapt and respond in real time to novel or polymorphic threats, leaving critical IoT infrastructure and sensitive data highly susceptible to exploitation. This project introduces the comprehensive design and development of an innovative AI-powered automated tool, meticulously engineered for seamless integration with heterogeneous IoT devices, specifically to address these emerging and dynamic cyber risks more effectively. By leveraging the continuous, real-time streams of operational and behavioral data generated from diverse IoT sensors and network endpoints, the system applies advanced deep learning—specifically Bidirectional Long Short-Term Memory (BiLSTM) networks. These networks are uniquely capable of analyzing intricate temporal sequences and learning complex behavioral baselines, allowing them to precisely detect subtle anomalies and assess potential vulnerabilities across interconnected networks. Unlike conventional static detection methods, BiLSTM models possess the intelligence to understand contextual patterns over time, identifying nuanced changes in device behavior or network traffic that could signify a nascent cyberattack or a compromised system.
DOI: http://doi.org/10.61463/ijset.vol.13.issue3.186
