International Journal of Computer Science Issues

More than a traditional journal...

The inadequacies inherent in the current defense mechanism of the mobile enterprise led to the development of new breed of security systems known as mobile intrusion detection system. The major worry of mobile / ubiquitous device users is the issue of data security since no mobile security application is 100% efficient. Existing studies conducted on android mobile security reveal that Android is the platform with the highest malware growth rate by the end of 2011 and that Global System for Mobile Communication -based Pivot Attacks, Mobile Botnets and Malicious Applications are the major security vulnerabilities compromising the confidentiality, integrity and availability of this mobile enterprise. In this paper, a SwarmDroid IDS is developed following a machine learning approach using Support Vector Machine. NSL-KDD dataset was used to test and evaluate the performance of the SwarmDroid IDS and compared with J48 and Random Forest which are state-of-the-art machine learning techniques for intrusion detection in mobiles. Particle Swarm Optimization was used for feature selection. The malware detection systems were simulated in a MATLAB environment. The SwarmDroid IDS was evaluated using detection time, true positive rate, false positive rate and detection accuracy as performance metrics. The result obtained from the evaluation revealed that SwarmDroid IDS outperforms J48 in terms of detection time and accuracy. Also, feature selection in Android application package files using particle swarm optimization technique plays a critical role in realizing high accuracy and low computational time complexity in SwarmDroid.