Challenges in Detecting Security Threats in WoT: A Systematic Literature Review
With the rapid development of the Internet of Things (IoT) and Web of Things (WoT), security issues have become increasingly prominent. In particular, the frequent occurrence of Denial of Service (DoS) attacks has made the security of WoT systems an urgent problem to be addressed. WoT achieves seamless connectivity between IoT devices and the internet by integrating IoT devices with web technologies, but this also introduces new security challenges. Due to the heterogeneity and openness of WoT devices, traditional security mechanisms struggle to cope with complex attack scenarios. Therefore, this paper aims to explore the security threats in WoT systems, especially the detection and defense mechanisms against DoS attacks, through a Systematic Literature Review (SLR), and analyze the application of Deep Learning (DL) techniques in this field.
Source of the Paper
This paper is co-authored by Ruhma Sardar, Tayyaba Anees, Ahmad Sami Al-Shamayleh, Erum Mehmood, Wajeeha Khalil, Adnan Akhunzada, and Fatema Sabeen Shaikh, from different research institutions. The paper was published in 2025 in the journal Artificial Intelligence Review, with the DOI 10.1007/s10462-025-11176-z.
Topic and Main Content of the Paper
Using the method of Systematic Literature Review, this paper reviews research from the past decade on security threats in WoT and IoT, particularly the detection and defense mechanisms against DoS attacks. The authors extracted key information from 80 relevant papers and proposed a classification of detection techniques based on deep learning. The main research questions addressed in this paper include: security threats in WoT environments, existing DoS detection mechanisms, the application of deep learning techniques in DoS detection, and future research directions.
1. Security Threats in WoT Environments
Due to the heterogeneity and openness of WoT devices, WoT systems face various security threats. These threats are mainly distributed across the four architectural layers of WoT: the Access Layer, Find Layer, Share Layer, and Compose Layer. Common threats include identity attacks, Man-in-the-Middle (MitM) attacks, data breaches, and DoS attacks. These threats not only affect system availability but may also lead to data privacy leaks and loss of system control.
2. Existing DoS Detection Mechanisms
Existing DoS detection mechanisms mainly include signature-based detection, anomaly-based detection, hybrid detection, and machine learning-based detection. Signature-based detection identifies threats by matching known attack patterns and is suitable for detecting known attacks. Anomaly-based detection identifies unknown attacks by analyzing abnormal behaviors in network traffic. Hybrid detection combines the advantages of both methods, enabling it to handle both known and unknown attacks. Additionally, machine learning-based detection techniques, particularly deep learning models such as Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM), have shown excellent performance in DoS detection but still require further optimization in terms of scalability and practical application.
3. Application of Deep Learning Techniques in DoS Detection
The application of deep learning techniques in DoS detection is mainly reflected in the following aspects: - Convolutional Neural Networks (CNN): CNN effectively identifies complex attack patterns by extracting features from network traffic. It shows high accuracy, especially when processing large-scale data. - Long Short-Term Memory Networks (LSTM): LSTM identifies periodic patterns in DoS attacks by memorizing the temporal information of network traffic. Research shows that LSTM achieves high accuracy in DoS detection, particularly when handling long time-series data. - Deep Belief Networks (DBN): DBN learns complex network traffic features through its multi-layer neural network structure, making it suitable for detecting multiple types of attacks.
4. Future Research Directions
Although deep learning techniques have shown excellent performance in DoS detection, there are still some challenges, such as model scalability, real-time performance, and adaptability. Future research should focus on optimizing deep learning models to improve their practical application in WoT environments. Additionally, lightweight detection mechanisms need to be developed to address the resource constraints of IoT devices.
Significance and Value of the Paper
Through a systematic literature review, this paper comprehensively reviews security threats in WoT environments and their detection mechanisms, particularly the application of deep learning techniques in DoS detection. The research not only provides theoretical support for the security of WoT systems but also points out directions for future research. By proposing a classification of detection techniques based on deep learning, this paper offers important references for developing efficient and scalable DoS detection mechanisms.
Highlights and Innovations
The highlights of this paper include: 1. Comprehensive Literature Review: Through a systematic analysis of 80 relevant papers, this paper provides a comprehensive summary of security threats and detection mechanisms in WoT environments. 2. Application of Deep Learning Techniques: This paper is the first to propose a classification of DoS detection techniques based on deep learning, providing new ideas for future research. 3. Proposal of Future Research Directions: In addition to summarizing existing research findings, this paper proposes future research directions, particularly in the optimization of deep learning models and the development of lightweight detection mechanisms.
Other Valuable Information
This paper also details the differences in security threats between WoT and IoT, especially the new vulnerabilities introduced by the integration of web technologies in WoT. Additionally, it explores security threats in different application domains (such as smart homes, healthcare, and smart cities), providing targeted recommendations for the security of WoT systems in various fields.
Through the research presented in this paper, readers can gain a deep understanding of the security challenges in WoT systems and their solutions, particularly the latest advancements in the application of deep learning techniques.