ACNN-BOT: An Ant Colony Inspired Feature Selection Approach for ANN Based Botnet Detection

With the growth and development of Internet and wireless communication, Internet-of-Things (IoT) has become a prominent technology for smart devices. In general, IoT systems are treated as badly secured because of always on nature of connectivity and high complexity of distributed computing. Because of this IoT systems have attracted different types of malicious attacks. Botnet is also one of the major types of malicious attacks of IoT systems. Early detection of IoT-Botnet can help to secure the network. In recent years, Machine Learning and Deep Learning have become popular tools for Botnet detection. The performance of these models mainly depends on the available features in the dataset. Selection of the most pertinent set of features may play an important role in order to improve the performance of Machine Learning and Deep Learning based detection models. In this paper, we proposed a novel feature selection approach inspired by Ant Colony optimization algorithm followed by Artificial Neural Network based IoT-Botnet detection. The proposed model has performed significantly well with an accuracy of 99.68% as well as in terms of precision and recall metrics. The proposed model also performed well in terms of feature selection with the improvement in the accuracy of 5% compared to complete feature set.