Implementation of ‘Smishing Detector’: An Efficient Model for Smishing Detection Using Neural Network

Abstract

Neural network creates a neuron-based network similar to the human nervous system to solve classification problems efficiently. The smishing problem is a binary classification problem in which attackers target smartphone users through text messages. As smishing is a remarkable cybersecurity issue that is troubling researchers and smartphone users these days. Addressing this security issue using the most efficient algorithm is the need of the hour. This manuscript presented an algorithm for the model proposed by authors in ‘Smishing Detector’ model and implemented it using Neural Network. The result obtained proves that the neural network is much efficient in detecting smishing problem. Neural Network outperformed other machine learning algorithms with a difference of 1.11%. Neural Network performed with the final accuracy of 97.40%. In this paper, system extracted the most efficient features of smishing SMS (Short Message Service) using the Neural Network. This manuscript also reported the accuracy shown by the system for each feature selected and implemented. It is evident from the implementation that each feature selected is most effective in smishing detection and URL (Uniform Resource Locator) feature is the most effective feature with an accuracy of 94%.

Detecting smishing attacks on smartphones: a comparative study between supervised and unsupervised learning techniques

“…From a literature review, we found that several research studies have been carried out in this direction, including the work of Sandhya Mishra and Devpriya Soni [2] on implementing a smishing attack detection system using neural networks. Their approach is bas features such as URLs to classify smishing SMS messages, which enabled them to obtain results with 97.40% accuracy for this work.…”

SmishGuard: Leveraging Machine Learning and Natural Language Processing for Smishing Detection

“…The experiment"s findings indicate a high level of accuracy, with the RF algorithm achieving an accuracy of 98.15% and the NB approach achieving an accuracy of 90.59%. Mishra et al [13] developed an efficient smishing detection system using an artificial neural network. The dataset utilized in the experiment consisted of 5858 messages, with 538 classifieds as smishing messages and 5320 as valid messages.…”

SmishSMS- The Latest Detection of SMS Phishing Trends

“…An accuracy of 97.40% is achieved by a neural network with a true positive rate value of 92.37% and a true negative rate value of 97.91%. [14] This paper considers four distinct strategies, focusing on deep learning, ensemble machine learning techniques and aberration detection in particular. The trials showed that the mentioned strategies still have a strong ability in showing the results with supreme precision of 99.95% for artificial neural network, anomaly detection having accuracy as 99.78%, XGBoost algorithm with 99.62% and 99.69% accuracy using random forest algorithm.…”