Vigistra-RealTime-25Class Dataset

Description The Vigistra-RealTime-25Class Dataset is a publicly available network intrusion detection dataset developed by Vigistra to support research on Network Intrusion Detection Systems (NIDS) using machine learning and artificial intelligence techniques. The dataset was collected from a controlled live network environment where attack scenarios were executed in real time against actively running network services. Unlike datasets generated from traffic replay or synthetic traces, the collected traffic represents live network communication observed during normal operation and controlled attack execution. Network traffic was monitored using Zeek, and the generated logs were processed into flow-based samples suitable for machine learning applications. A total of 37 flow-based features were extracted to characterize network communication behavior. The dataset contains 1,431,737 network flows, including 1,057,737 benign flows and 374,000 attack flows distributed across 24 attack classes and 1 benign class. The attack scenarios cover multiple categories, including: Denial of Service (DoS) Distributed Denial of Service (DDoS) Brute Force attacks Web application attacks Network scanning Vulnerability scanning Data exfiltration Fuzzing Backdoor activities Traffic labels were generated automatically based on the controlled attack execution process, reducing the possibility of manual labeling errors and improving label consistency. To protect privacy, source and destination IP addresses have been anonymized using hashing techniques before publication. The dataset is intended for research in multi-class intrusion detection, machine learning model evaluation, feature selection, cybersecurity benchmarking, and related network security studies. This release includes: Five CSV dataset files Feature description (Feature_Description.csv) Dataset documentation (Dataset_Description.pdf) Dataset analysis script README documentation DOI: https://doi.org/10.5281/zenodo.21704609