A Clever Approach to Develop an Efficient Deep Neural Network Based IDS for Cloud Environments Using a Self-Adaptive Genetic Algorithm

Nowadays, Cloud Computing is one of the fastest growing and most used computing paradigms in the IT field. It is a computational platform that integrates massive computing, storage and network resources into a unified pool of resources, and offers them online over Internet to customers in an on-demand and pay-per-use fashion with least involvement of the cloud service provider. This new archetype characterized by big data and distributed technology uses such technology as multi-tenancy and virtualization, which brings along vulnerabilities, sharing risks and lead to different matters related to security and privacy in cloud computing (CC). Therefore, it is essential to create an efficient intrusion detection system to detect intruders and suspicious activities in and around the CC environment by monitoring network traffic, while maintaining performance and service quality. In this work, we propose a clever approach using a self-adaptive genetic algorithm (SAGA) to build automatically a Deep Neural Network (DNN) based Anomaly Network Intrusion Detection System (ANIDS). SAGA is a variant of standard Genetic Algorithm (GA), which is developed based on GA improved through an Adaptive Mutation Algorithm (AMA). Our method consists of using SAGA with the purpose of looking for the optimal or near optimal combination of most relevant values of the parameters included in building of DNN based IDS or impacting its performance, like feature selection, data normalization, architecture of DNN, activation function, learning rate and Momentum term, which ensure high detection rate, high accuracy and low false alarm rate. CloudSim 4.0 simulator platform and Kyoto 2006+ dataset version 2015 were employed for simulation and validation of the proposed system. The experimental results obtained demonstrate that in comparison to several traditional and recent approaches, our proposed IDS achieves higher detection rate and lower false positive rate.