SMISHFILTER : AN ENSEMBLE MACHINE LEARNING BASED MODEL TO DETECT SMS PHISHING ATTACKS

Abstract

In the current technological era, people are attracted to electronic gadgets like laptops and mobile phones, mainly smart phones and tablets, which have become their primary source of entertainment in the current virtual world. People are more attached to their mobile phones due to the easy availability of mobile internet and pocket-friendliness, which exposes various security threats like phishing, Smishing, and information leaks due to malicious app installation. There are multiple mediums through which phishing attacks are generally carried out, like fake websites, Emails, or SMS. Generally, people prefer SMS over mail because it is simple at the same time and doesn’t need an internet connection. We can’t ignore the fact that the cost of SMS has also decreased a lot; hence the use of SMS service has increased. The rise in the usage of SMS services made attackers use it as the medium. Here we have proposed an efficient approach for detecting SMS-based phishing called Smishing. To differentiate between Smishing and legitimate messages, 75 different features are used initially. The results of our experiments show that our approach gives an accuracy of 98.93 % using Ensemble machine Learning Techniques. We integrate an android malware detector, which provides security against information leaks due to APK installation with or without user consent which spread through SMS. Also, we are able to select optimal feature set using various correlation algorithms, which select 18 features and give an accuracy of 98.81%, which is better than earlier work that used a correlation algorithm.