Isolation of DDoS Attacks and Flash Events in Internet Traffic Using Deep Learning Techniques

Abstract

The adoption of network function visualization (NFV) and software-defined radio (SDN) has created a tremendous increase in Internet traffic due to flexibility brought in the network layer. An increase in traffic flowing through the network poses a security threat that becomes tricky to detect and hence selects an appropriate mitigation strategy. Under such a scenario occurrence of the distributed denial of service (DDoS) and flash events (FEs) affect the target servers and interrupt services. Isolating the attacks is the first step before selecting an appropriate mitigation technique. However, detecting and isolating the DDoS attacks from FEs when happening simultaneously is a challenge that has attracted the attention of many researchers. This study proposes a deep learning framework to detect the FEs and DDoS attacks occurring simultaneously in the network and isolates one from the other. This step is crucial in designing appropriate mechanisms to enhance network resilience against such cyber threats. The experiments indicate that the proposed model possesses a high accuracy level in detecting and isolating DDoS attacks and FEs in networked systems.