Quantum-Inspired Feature Learning and Parameter-Efficient Neural Architectures for IoT Intrusion Detection: A Comprehensive Review and Future Research Roadmap
Jul 2026· 2026 6th International Conference on Inventive Computation and Information Technologies (ICICIT)· pp. 1026-1033· 0 citations· 20 references
Abstract
Rapid growth in the Internet of Things (IoT) networks means that almost limitless connectivity of intelligent devices and advanced technologies in the domains of healthcare, transport, industrial processes, and smart city automation are possible. However, the mass rollout of IoT devices with limited resources has exposed IoT networks to numerous cyber threat vectors including botnets, distributed denial-of-service (DDoS) attacks, the spread of malware, and unauthorized intrusions to computer networks. This has prompted the evolution of Intrusion Detection Systems (IDSs) from traditional machine learning (ML) techniques to various advanced deep learning techniques used for the detection and analysis of sophisticated cyberattack techniques on network traffic. Within this context, deep neural networks, convolutional neural networks, recurrent neural networks, and various autoencoder models, have all produced significant results with respect to attack detection; however, practical implementations have been constrained by a considerable amount of time and space resources. This has led to a focused interest in quantum-inspired learning approaches, which improve feature distinguishability for normal and attack traffic through the application of high dimensional probabilistic mappings. In addition, frameworks that utilize parameter-efficient design based on tensor decomposition, low-rank, and Kronecker factorizations, have reduced the time and space resources needed for attack detection. A comprehensive overview of IoT intrusion detection research, including benchmark datasets and ML and deep learning models, as well as quantum-inspired and structured neural approaches is provided in this paper. Additionally, current approaches are evaluated on their strengths and weaknesses, and major research gaps are noted. Lastly, the focus is on the development of the next generation of IoT security that is intelligent, scalable, and able to provide real-time protection.
An intelligent cyberattack detection system that applies machine learning and deep learning techniques to classify network traffic as either normal or malicious, and demonstrates the potential of machine learningbased intrusion detection systems in improving network security and supporting the protection of modern smart environments.
KADADHARAPU ANUPRIYA, Dr.S.SWATHI RAO· International Journal of Eng...· 0 citations
The framework introduces CNN–BiLSTM deep learning networks to represent traffic in a spatiotemporal manner and adopts ensemble machine learning classifiers to enhance the robustness of traffic detection and its interpretability, to enhance the robustness of traffic detection and its interpretability.
Ramesh N. S. V. S. C. Sripada, A. Bhavani, Kiran B. Malagi et al.· Discover Computing· 0 citations
The rapid proliferation of Internet of Things (IoT) devices across critical domains including healthcare, smart cities, industrial control systems, and intelligent transportation has fundamentally transformed the cybersecurity threat landscape. The inherent characteristics of IoT environments, namely resource-constrained devices, heterogeneous architectures, and large-scale deployment, render traditional Intrusion Detection Systems (IDS) inadequate for the sophisticated and evolving attack vectors targeting these networks. Deep learning (DL) has emerged as a compelling paradigm for next-generation IoT IDS, offering automated feature extraction, temporal pattern recognition, and adaptive threat detection capabilities that address the limitations of conventional approaches.
This paper provides a thorough and systematic review of the existing DL methods for IoT intrusion detection. The paper explore the IoT architectural paradigms, outline a four layered taxonomy for types of IoT attacks across its three primary layers Perception, Network and Application as well as Adversarial Machine Learning attacks, and systematically review seven classes of DL architectures Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRU), Autoencoders, Generative Adversarial Networks (GAN), models based on Transformer architecture and Federated Learning frameworks. In a comparative review of forty peer-reviewed studies, we demonstrate that hybrid DL models provide excellent detection performance (99-100% classification accuracy on benchmark datasets) as well as practical viability for deployment with privacy-preserving Federated Learning for large-scale data. The study additionally highlights five enduring challenges class imbalance, adversarial vulnerability, zero-day detection limitations, computational constraints and the absence of standardized benchmarking protocols that together account for the gap between performance benchmarks and real-world deployment efficacy. It outlines future research avenues targeting on five key axes with a particular focus in the integration of Explainable AI (XAI), lightweight edge-deployable architectures, and adversarial robustness mechanisms. This survey identifies a structured reference to advance the state of IoT intrusion detection from research to operationally viable and deployable systems.
Mohammed Gharkan, Mustafa I. Hussien Al-Janabi, Obaid Salim· Al-Noor Journal of Engineeri...· 0 citations
The rapid expansion of the Internet of Things (IoT) has raised additional concerns about security, and there was a major risk of Distributed Denial-of-Service (DDoS) attacks because the IoT devices have limited computation, memory, and energy capabilities. Traditional intrusion detection methods, which are at times contrived to support a high capacity, are incompetent at these limitations, delaying detections, having too many false alarms, and also compromising the system performance. This study offers a resource-efficient, adaptive machine learning system that was suitable to be used in the operation of DDoS attacks in resource-confined IoT settings. The technique combines the hybrid feature selection algorithms based on mutual information gain and recursive feature elimination to construct a more compact and high-utility feature set together with the optimization of the lightweight classifiers, including stochastic gradient descent and shallow decision trees. The concept drift was solved by an online incremental learning mechanism that guarantees long-term trend detection over time against changing patterns of attacks. The evaluation of the benchmark datasets (CICDDoS2019, BoT-IoT, TON_IoT) using experimental evaluation on a heterogeneous testbed IoT and assessing both security metrics and resource efficiency was researched. The model suggested had a precision of 0.973, a recall of 0.959, an F1-score of 0.966, and an average decrease of malicious traffic by 93 percent at the expense of legitimacy throughput. Latency was decreased to 2.6 seconds when detecting high-intensity attacks, and the CPU and memory usage continued to be less than 35 percent and 70 percent of the device capacity, respectively. A better result in terms of accuracy, response time, false positive rates, and not using resource budgets was witnessed when compared to baseline models through comparative analysis. The results verify the framework's ability to provide low latency and correct DDoS mitigation directly on the IoT devices, which can be considered a feasible solution to achieve resilience improvement of critical IoT deployments in health care, industrial automation, and smart cities.
Selvi T, Jayaganesh J· International journal of com...· 0 citations
A taxonomy of IDS architectures and a survey of recently proposed networks, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), LSTMs, GRUs, Autoencoders, GANs, Transformer-based, as well as hybrids are developed.
Sajith K. V., Gripsy Paul, Bhagavant Deshpande et al.· Journal of Network and Infor...· 0 citations
The rapid growth of digital communication technologies, cloud computing, and Internet of Things (IoT) devices has
increased both the frequency and sophistication of cyber-attacks, making effective intrusion detection an essential component of
modern cybersecurity systems. Traditional signature-based intrusion detection systems (IDS) are effective against known attacks
but fail to detect previously unseen or evolving threats. This study investigates the application of deep learning models for binary
network intrusion detection using the NSL-KDD benchmark dataset. Three standalone architectures, Convolutional Neural
Networks (CNN), Long Short-Term Memory (LSTM) networks, and Deep Neural Networks (DNN), are implemented and
evaluated, alongside a CNN-LSTM Hybrid model that integrates spatial and sequential learning, and a DNN-LSTM Ensemble
model that combines independently trained DNN and LSTM predictions through weighted averaging. Following data cleaning,
categorical encoding, normalization, and Random Forest-based feature selection (41 features reduced to 20), all models were
trained and evaluated under identical conditions using Accuracy, Precision, Recall, F1-Score, ROC-AUC, training time, and
inference time. The standalone DNN model achieved the best overall performance, with 80.98% accuracy, 97.08% precision,
68.66% recall, 80.43% F1-score, and 96.11% ROC-AUC, while also requiring the shortest training time (39.69 s). The CNNLSTM Hybrid model attained the highest precision (97.23%) but did not outperform the standalone architectures overall, and
the DNN-LSTM Ensemble produced balanced but not superior results. These findings indicate that carefully designed
standalone architectures can match or exceed the performance of more complex hybrid and ensemble models for binary
intrusion detection, while incurring substantially lower computational cost. The study contributes a controlled, commonframework comparison of five deep learning architectures and provides practical guidance for selecting computationally
efficient models for anomaly-based intrusion detection.
Ketki Naik, Sanjeev Ghosh· International Journal for Re...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
