A novel approach to defend multimedia flash crowd in cloud environment

Cloud computing is an intelligent integration of distributed computing, hardware virtualization techniques, automated data center techniques and Internet technologies. Due to its appealing features, it has become most prevailing computing platform. Since, a large number of customers are moving towards cloud, attackers are also more interested in attacking cloud services. Distributed Denial of Service (DDoS) attack is one of the most popular methods to disrupt the services of a cloud platform hosting multimedia services. Modern day attackers use botnets to perform variety of DDoS attacks. With the advancement in the technology, bots are now capable to simulate the DDoS attacks as flash crowd events. During a flash crowd event, requests are sent by legitimate users; therefore these requests should not be denied and the server should be able to ensure user's QoE during a flash crowd event. Based on our study of botnets, flash crowd and DDoS attacks, in this paper, we propose a flow confidence based discrimination algorithm to distinguish between flash crowd event and DDoS attack. Moreover, we have given an effective, efficient and economical approach to ensure user's QoE during flash crowd events. We have performed various experiments using benchmark datasets to support our theoretical claims which also determine the efficiency and effectiveness of the proposed approach in real world scenario.