Besides the developing threat of cyberattacks, cybersecurity is one of the most significant Internet of Things (IoT) regions. While the IoT has formed a novel model where a network of devices and machines is efficient in collaborating and communicating with each other, it is a novel process invention in enterprises. The role of cybersecurity is to mitigate risks for institutions and users by safeguarding data confidentiality across networks. The growing tools and technologies for cybersecurity improve safety in IoT systems. AI is helpful in improving cybersecurity by providing real-time information for faster threat detection, rapid responses, and smarter decisions. Moreover, integrating blockchain (BC) with IoT illustrates promise but encounters threats like performance issues, security vulnerabilities, and scalability limits. Still, BC plays a key part in protecting low-energy IoT devices. In this study, I proposed a novel approach using an Attention Mechanism-Based Recurrent Neural Network and Dimensionality Reduction for Cyber Threat Detection (AMRNN-DRCTD) model. The main goal of the proposed AMRNN-DRCTD model was to enhance the detection system for cyberattacks in IoT networks. I considered possible security breaches in BC and their influence on network processes. At the initial stage, the data normalization applied zero-mean normalization to alter data into a consistent setup. The feature selection process employed the chaotic and terminal strategy-based butterfly optimization algorithm (CTBOA). Furthermore, the proposed AMRNN-DRCTD model utilized the hybrid convolutional neural network and bi-directional long short-term memory with an attention mechanism (CNN-BiLSTM-AM) technique for the classification process. Finally, the Honey Badger Algorithm (HBA)-based hyperparameter selection range was accomplished to optimize the detection outcomes of the CNN-BiLSTM-AM technique. The experimental evaluation of the AMRNN-DRCTD methodology was examined under the BoT-IoT dataset. The performance validation of the AMRNN-DRCTD methodology highlighted a superior accuracy output of 99.28% over existing approaches.

With the widespread use of Internet, Internet of Things (IoT) devices have exponentially increased. These devices become vulnerable to malware attacks with the enormous amount of data on IoT devices; as a result, malware detection becomes a major problem in IoT devices. A reliable and effective mechanism is essential for malware detection. In recent years, research workers have developed various techniques for the complex detection of malware, but accurate detection continues to be a problem. Ransomware attacks pose major security risks to corporate and personal information and data. The owners of computer-based resources can be influenced by monetary losses, reputational damage, and privacy and verification violations due to successful assaults of ransomware. Therefore, there is a need to swiftly and accurately detect the ransomware. With this motivation, the study designs an Ebola optimization search algorithm for enhanced deep learning-based ransomware detection (EBSAEDL-RD) technique in IoT security. The purpose of the EBSAEDL-RD method is to recognize and classify the ransomware to achieve security in the IoT platform. To accomplish this, the EBSAEDL-RD technique employs min-max normalization to scale the input data into a useful format. Also, the EBSAEDL-RD technique makes use of the EBSA technique to select an optimum set of features. Meanwhile, the classification of ransomware takes place using the bidirectional gated recurrent unit (BiGRU) model. Then, the sparrow search algorithm (SSA) can be applied for optimum hyperparameter selection of the BiGRU model. The wide-ranging experiments of the EBSAEDL-RD approach are performed on benchmark data. The obtained results highlighted that the EBSAEDL-RD algorithm reaches better performance over other models on IoT security.

The internet is the most effective means of communication in the modern world. Therefore, cyber-attacks are becoming more frequent, and their consequences are becoming increasingly severe. Distributed denial of service (DDoS) is one of the five most effective and costly cyberattacks. DDoS attacks are the most prevalent and expensive in today's evolving cybersecurity landscape. However, their ability to disrupt network services causes significant financial losses and has become an effective means of DDoS detection and prevention, both of which are essential for organisations. Network monitoring and control systems have found it challenging to recognise the numerous classes of denial of service (DoS) and DDoS attacks, as they all work exclusively. Therefore, an effective model is needed for attack detection. A previous study has established that shallow and deep learning (DL) methods are vital for identifying DDoS threats; however, there is a lack of research on time-based features and classification across numerous DDoS threat categories. This manuscript introduces an ensemble learning model integrated with two-tier heuristic optimisation techniques for effective cyber defence (ELMT2HO-ECD) methodology. The primary purpose of the ELMT2HO-ECD methodology is to provide a robust solution for detecting and mitigating DDoS attacks in real time. Initially, the ELMT2HO-ECD approach applies mean normalisation to the data to measure the feature within a specified range. Furthermore, the mountain gazelle optimiser (MGO) approach is utilised for feature extraction. For DDoS attack detection, ensemble DL models, namely convolutional long short‐term memory (ConvLSTM), Wasserstein autoencoder (WAE), and temporal convolutional networks (TCN), are employed. To further enhance the performance of the three ensemble models, hyperparameter tuning is performed using the improved pufferfish optimisation algorithm (IPOA), which optimises the models' parameters to achieve higher accuracy. The ELMT2HO-ECD model is evaluated on the CICIDS2017, CICIDS2018, and CICIDS2019 datasets. Validation of the performance of the ELMT2HO-ECD model demonstrated superior accuracy of 98.93%, 98.43%, and 99.23% compared with existing techniques.