By: Vanna karthik; Vel Tech University, Chennai, India
Abstract
Smishing, a type of phishing that is carried out via SMS messages, presents serious security risks since it deceives users into disclosing private information. The changing strategies of cybercriminals have made traditional methods of smishing detection ineffective. This research investigates the benefits of improving the smishing detection system using deep learning techniques. We showcase the ways through which deep learning in pattern recognition and parsing natural language can easily catch such attempts at smishing. Our work shows that deep learning enhances precision and efficiency in the detection of smishing, thus opening a more intelligent approach toward this problem of cybersecurity.
Introduction
Due to the increasing number of mobile devices, SMS-based attacks have also increased, including smishing, whereby criminals send fake messages to the victim to install malware or steal personal data. Traditional methods for smishing detection, such as rule-based systems and algorithmic methods, can hardly keep pace with the ever-changing attacking strategies. Deep learning represents one of the branches of machine learning that have complex pattern recognition and self-adjusting capabilities with new data. Some deep learning models in smishing detection were presented in this paper and the benefits of using them compared to traditional techniques were underlined.
Literature review
Recent research has highlighted the weakness of conventional smishing detection techniques. Rule-based systems, for example, need to be updated frequently and are not very good at generalizing to new kinds of threats. [1]developed a hybrid CNN-LSTM model for SMS spam detection in Arabic and English messages. While their work focused on spam detection rather than smishing specifically, the hybrid model they proposed may offer advantages for capturing complex patterns and relationships in smishing messages.
One of the popular usages in recent years has been smishing detection with deep learning algorithms. CNNs and RNNs are two important variants of deep learning that have shown much promise in the detection of intricate patterns and relationships in textual data. These algorithms can learn to automatically extract important information and generate precise predictions by being trained on big datasets of tagged SMS messages[2].
It consists of data pre-processing, local model training on client devices, and model aggregation. This can classify newly received SMS messages accurately as spam or real. Interestingly, the design protects data privacy by storing data on client devices and allows model training to be done cooperatively without exchanging data [3].
While machine learning approaches are more flexible, they often rely on feature engineering, which is time-consuming and less effective against novel smishing strategies. The deep learning models, especially those with RNNs and CNNs, have achieved great performance in various NLP tasks, including spam detection. These models can automatically learn features from raw data, making them well-suited for identifying the nuanced and evolving characteristics of smishing messages[4].
Methodology
In this paper, we used a dataset of labeled SMS texts to create a deep learning system for smishing detection. To examine the messages’ text content, we used CNNs, LSTM and RNNs. To guarantee the robustness of the models, the dataset was divided into subsets for testing, validation, and training. Tokenization, embedding, and normalizing were preprocessing processes. The models’ efficacy was assessed using performance indicators like accuracy, and precision.
Conclusion
Our study shows that, compared to traditional methods, deep learning models provide significant advantages in smishing detection. These algorithms can adapt to new and evolving smishing methods by dynamically learning characteristics from data, which raises the detection rate and lowers false positives. The findings suggest that smishing detection systems that embed deep learning might offer a more intelligent and robust defense against phishing attacks that rely on SMS.
References
- M. K. Mehmood, H. Arshad, M. Alawida, and A. Mehmood, “Enhancing Smishing Detection: A Deep Learning Approach for Improved Accuracy and Reduced False Positives,” IEEE Access, vol. 12, pp. 137176–137193, 2024, doi: 10.1109/ACCESS.2024.3463871.
- K. Gowri and S. Brindha, “ADVANCEMENTS IN SMISHING DETECTION: A NOVEL APPROACH UTILIZING GRAPH NEURAL NETWORKS,” vol. 24, no. 1000, 2024.
- M. A. Remmide, F. Boumahdi, B. Ilhem, and N. Boustia, “A privacy-preserving approach for detecting smishing attacks using federated deep learning,” Int. J. Inf. Technol., Aug. 2024, doi: 10.1007/s41870-024-02144-x.
- M. R. A. Saidat, S. Y. Yerima, and K. Shaalan, “Advancements of SMS Spam Detection: A Comprehensive Survey of NLP and ML Techniques,” Procedia Comput. Sci., vol. 244, pp. 248–259, 2024, doi: 10.1016/j.procs.2024.10.198.
- Sharma, A., Gupta, B. B., Singh, A. K., & Saraswat, V. K. (2023). Advanced persistent threats (apt): evolution, anatomy, attribution and countermeasures. Journal of Ambient Intelligence and Humanized Computing, 14(7), 9355-9381.
- Xu, M., Peng, J., Gupta, B. B., Kang, J., Xiong, Z., Li, Z., & Abd El-Latif, A. A. (2021). Multiagent federated reinforcement learning for secure incentive mechanism in intelligent cyber–physical systems. IEEE Internet of Things Journal, 9(22), 22095-22108.
- Zhou, Z., Li, Y., Li, J., Yu, K., Kou, G., Wang, M., & Gupta, B. B. (2022). Gan-siamese network for cross-domain vehicle re-identification in intelligent transport systems. IEEE transactions on network science and engineering, 10(5), 2779-2790.
- KEE S.N. (2024) Blockchain for Decentralized IoT Security: Protection Against Phishing and Data Breaches, Insights2Techinfo, pp.1
Cite As
Karthik V. (2025) The Deep Learning Advantages : Smishing Detection made Smarter, Insights2techinfo pp.1