Deep Learning – Based Network Intrusion Detection for Internet of Things (IoT) Using Bidirectional Long Short-Term Memory (BLSTM): A Review

@article{1720213,
      author = {Aleng E. A., Sarjiyus O., Yusufu G.},
      title = {Deep Learning – Based Network Intrusion Detection for Internet of Things (IoT) Using Bidirectional Long Short-Term Memory (BLSTM): A Review},
      journal = {Iconic Research And Engineering Journals},
      year = {2026},
      volume = {10},
      number = {1},
      pages = {3707-3721},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1720213.pdf},
      abstract = {Deep Learning works by copying how the human brain operates. To get these features, the deep learning method uses strong neural network algorithms, like clustering algorithms, Bayesian algorithms, and artificial neural network algorithms. Deep learning algorithms are very powerful in terms of processing power, which makes them better suited for handling complicated and mixed data sets from IoT devices than traditional machine learning methods. The use of deep learning in the Internet of Things, especially when it comes to securing IoT networks, is still in the early stages of research and has a lot of promise for identifying security threats within IoT systems. Even though the structures of recurrent neural networks can be pretty complicated, adjusting the hyper parameters can help make them work well for IoT applications. This idea encourages using deep learning techniques to improve the safety of IoT networks. The fast growth of the Internet of Things (IoT) has created a large area where cyber threats can spread, putting important systems and personal information at risk from more advanced attacks. Traditional network intrusion detection systems (NIDS) use fixed signatures and require manual setup of features. Because of this, they can't find new types of attacks, like zero-day exploits and changing malware, which are common now. The different types of data from IoT devices, which use various communication methods like MQTT and CoAP, move quickly and vary a lot. This makes the data patterns complex and not straight forward, making it hard for regular systems to properly organize or understand them. There is a big need for a detection system that can: look at network traffic as a ongoing, two-way flow to get a complete picture. Lower the number of false alerts that cause "alert fatigue" for network administrators. Work well against the changing and urgent attack methods that are specific to IoT systems. The method to be used for collecting data is the secondary approach. Because I will use literature reviews from other authors to fill in the gap that has been identified. Research shows that there is a big need to develop good Deep Learning models that can detect attacks in data sets. The NSL-KDD dataset will be studied and used to train four different deep learning methods.},
      keywords = {Artificial Neural Network, Attack Vectors, Cyber Threats, Deep Learning (DL), Internet of Things (IoT), Network Intrusion Detecting System (NIDS) and Polymorphic Malware.},
      month = {July},
  }