An Efficient Technique for Identifying Distributed Denial of Service Active Assaults Using Deep Neural Networks Based on the Adaptive System Intelligence Paradigm

Abstract

A collection of interconnected devices that exchange data online is known as the Internet of Things. The IoT environment's diverse components make the distributed denial-of-service attack a security risk. One of the most important tasks in creating a smarter environment for end users is detecting DDoS attacks in the Internet of Things. A new version of the optimized Elman recurrent neural network (ERNN) is proposed to detect DDoS active attacks in Internet of Things scenarios. The proposed detection approach optimizes the weight and bias of ERNN (ABCO-ERNN) using a novel adaptive bacterial colony optimization (ABCO) technique. The ABCO algorithm uses an adaptable step size to increase the BCO's capacity for both exploration and exploitation. The four datasets, BoT-IoT, CIC-IDS2017, CIC-DDoS2019, and IoTID20, are used to compare performance, and five distinct performance measures, including accuracy, precision, sensitivity, specificity-ty, and f-measure, are considered. When compared to previous literature algorithms, the proposed ABCO-ERNN detection approach produced a high detection rate according to the experimental results.