An efficient DDoS attack detection mechanism based on distributed self organizing map

Abstract

Even in the era of 5G and beyond networks, we are struggling against an old problem, DoS (Denial of Service) attack which has arisen since the beginning of network. Since it is still the biggest threat of all cyber-attacks and it is growing more complex, there has been a lot of effort to mitigate DDoS attacks. As one of the various approaches, machine learning techniques are applied to detect attacks. Self-Organizing Map (SOM), one of unsupervised learning mechanisms, has been used as a tool to detect DoS attacks. However, existing SOM-based approaches have the potential drawbacks, the limited detection throughput and vulnerability to DoS attack. Therefore, this paper proposes a new concept of SOM, Distributed SOM (DSOM), and a novel DDoS attack detection mechanism based on DSOM. Through the simulation and experiment results, we show the proposed mechanism provides a feasible and efficient detection. © Research India Publications.