Downloads: 1 | Views: 74 | Weekly Hits: ⮙1 | Monthly Hits: ⮙1

Research Paper | Computer Science and Information Technology | Iraq | Volume 13 Issue 11, November 2024 | Popularity: 5 / 10

Rate This Article!      

Detection and Prevention of Distributed Denial of Service (DDoS) Attacks using Metaheuristic and Machine Learning Techniques

Ameer Sameer Hamood Mohammed Ali

Abstract: The Internet of Things (IoTs) is vulnerable to DDoS attacks, which provide a significant risk to many web-based networks. The intruder's capacity to manage the potential of diverse collaborative gadgets in order to initiate an attack further complicates its administration. The level of complexity can be further heightened when several attackers endeavor to overwhelm a device through a sustained attack. In order to mitigate and safeguard against contemporary DDoS attacks, several efficacious and robust methodologies have been employed within scholarly discourse. These methodologies encompass the utilization of data mining and artificial intelligence within the realm of Intrusion Detection System (IDS). However, it is important to acknowledge that these methodologies are not without their limits. In order to address the current constraints. In this paper, we propose DDoS attack detection and preventing approach using Hybrid model integrated Particle Swarm Optimization (PSO) metaheuristic algorithm and Machine Learning techniques as PSO-ML model. The proposed PSO in IoT network is used for optimizing performance, reducing energy consumption, load balancing, and ensuring scalability, it making IoT suitable for complex and multidimensional optimization problems often encountered in IoT resource management. It evaluates the fitness of each particle by training a DDoS attack detection model with machine learning classifier on the selected features and measuring its performance. PSO-ML model is capable of distinguishing between normal and malicious network traffic. The results showed that the Hybrid PSO-ML DDoS defense system is useful for automating the feature selection process, enhancing the efficiency of DDoS attack detection, high accuracy of DDoS attack detection, best accuracy of UNSW-NB15 dataset is 99.64 % of MLP, CICIDS2017 Dataset is 99.53% of RF, DDOS attack SDN Dataset is 99.54 %, KDDCUP99 Dataset is 97.52 % of RF. Besides, the Average processing time is 41.651 seconds, 149.766 seconds, average packet delivery ratio is 99.65%, 17.35%, average network utilization is 9.791 KB, 0.812 KB, resource utilization 32.061%, 4.572% and Average throughput is 23446.861 KB, 3374.847 KB of PSO-ML Model and Without Optimization within DDos attack respectively.

Keywords: Internet of Things (IoTs), Distributed Denial of Service (DDoS), Machine Learning, PSO metaheuristic algorithm

Edition: Volume 13 Issue 11, November 2024

Pages: 633 - 642

DOI: https://www.doi.org/10.21275/SR241107015558

Make Sure to Disable the Pop-Up Blocker of Web Browser

Click to Cite this Article

Similar Articles