An efficient DDoS attack detection method in IoT based on an optimized recurrent neural network using hybrid swarm intelligence methods

Abstract

The recent growth of the Internet of Things (IoT) has worsened security concerns, and one of the most common threats today is the Distributed Denial of Service (DDoS) protection because of the modest computational capabilities and the heterogeneity of the devices comprising the IoT. Conventional intrusion detection systems frequently fail to distinguish between legitimate and malicious flows very well, where the former results in high false alarms and low adaptability to changing patterns of attacks. This paper has suggested an effective DDoS attack detection algorithm based on an optimized Elman Recurrent Neural Network (ERNN) with an improved algorithm, the Improved Grey Wolf Optimization (IGWO) algorithm, to overcome these limitations. The ERNN is utilized to learn the time dependencies in the IoT traffic, and the IGWO is utilized to optimize the weights, biases, and hyperparameters of the framework,including adaptive convergence control, opposition-based learning, and enhanced search strategies. Experimental analyses of various datasets of IoT devices, such as Fridge, Garage, GPS, Modbus, Light Motion, and Thermostat. The IGWO-ERNN is more accurate, precise, and recalls specific information, and has a higher F-measure, faster convergence, and lower misclassification. These findings demonstrate the strength and effectiveness of the IGWO-ERNN architecture, which would be suitable in real-time and resource-constrained IoT settings and can contribute significantly to countering DDoS attacks.