SDN-based Deep Anomaly Detection for Securing Cloud Gaming Servers

Abstract

Despite recent advances in cloud computing, users and organizations have always feared for the security of cloud environments. On the other hand, there is a concern on the part of cloud service providers, since all the cloud infrastructure shares sensitive data on the Internet. For this reason, an in-depth study to diagnose network anomalies seems logical, because with a precise approach, the risks of infiltration can be reduced. In this paper, we have used Software Defined Network (SDN) to implement game streaming in order to achieve our test penetration. Furthermore, we built our SDN-based database by performing a greedy approach. For this job, during multiple game streaming, three attackers infiltrate the cloud game infrastructure in a variety of ways to make the access of the gamer and the game server out of reach. By using the data from this event, which are stored in the controller, we have created a Neural Network (NN) to assess and diagnose abnormalities. Numerical results show that our controller can be effective in detecting anomalies with very little error.