KS-DDoS: Kafka streams-based classification approach for DDoS attacks.

Area of Science:

• Computer Science
• Cybersecurity
• Network Security

Background:

• Distributed Denial of Service (DDoS) attacks pose a significant threat to internet-based systems, causing service disruptions and financial losses.
• Existing DDoS mitigation techniques struggle to keep pace with the increasing frequency and sophistication of attacks.

Purpose of the Study:

• To propose and validate a novel distributed classification approach, KS-DDoS, for real-time detection and classification of DDoS attacks.
• To leverage Apache Kafka Streams and Hadoop for scalable and efficient DDoS threat identification.

Main Methods:

• Designed distributed machine learning models on a Hadoop cluster using data from the Hadoop Distributed File System (HDFS).
• Deployed a real-time classification model on a Kafka Streams cluster to categorize network traffic into nine classes.
• Stored discriminative features and outcomes in HDFS for continuous model improvement.

Main Results:

• The proposed KS-DDoS approach demonstrated efficient real-time classification of incoming network traces.
• Achieved a classification accuracy of at least 80% for DDoS attack detection.
• Validated the effectiveness through a distributed processing framework-based experimental environment.

Conclusions:

• The KS-DDoS approach offers an effective solution for real-time distributed classification of DDoS attacks.
• The integration of Kafka Streams and Hadoop enables scalable and robust network security.
• The system provides a foundation for adaptive and continuously improving DDoS defense mechanisms.