Dataset for DoS Detection Dataset: A Benchmark for AI-Driven Intrusion Detection and DoS Attack Analysis
Year: 2024
Type: article
Abstract: The "DOS ATTACK" dataset, which consists of network traffic data specially prepared to support the research effort into finding the possibility to trace DoS attacks using artificial intelligence and machine learning-based techniques, has more than 820,000 entries. Labeled instances of attack and normal traffic represent this dataset, which are highly useful for training and evaluation of intrusion detection systems. Some of the notable features include source and destination IP addresses, protocol type, packet size, and TCP-specific flags like SYN and ACK, which are actually quite crucial to observe patterns of attacks. The dataset is fairly balanced as about 17 percent of entries fall under the category of attack instances, which gives very realistic settings for evaluating the model. This dataset may also be important for benchmarking AIbased network intrusion detection models, especially for recognition of DoS attacks and general knowledge about network behavior under attack conditions. Salient patterns of distribution in the SYN and ACK flags, packet sizes, and temporal characteristics shown in our analysis could constitute a foundation for more advanced intrusion detection research
Author: Shihab Chilwan
Institution: Vishwakarma University
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4405602807