LightFlow-DDoS

The LightFlow-DDoS dataset is a large-scale, flow-based network traffic dataset designed for detecting Distributed Denial-of-Service (DDoS) attacks. It consists of 8,45,373 network flow records extracted using CICFlowMeter-style feature engineering and comprises 40 statistical flow-level features related to packet size, traffic volume, flow duration, inter-arrival time and transmission rates. The dataset includes essential network identifiers such as source and destination IP addresses, ports, protocols, and timestamps, along with derived features such as flow bytes per second, packet length statistics, and forward packet dynamics. Traffic is labelled using a dual-labelling scheme, where a categorical label indicates the attack type and a binary label distinguishes between benign and malicious traffic, enabling both binary and attack-specific classification tasks. The dataset primarily focuses on LDAP-based DDoS traffic, making it computationally efficient and well-suited for machine learning–based DDoS detection, real-time cyber–physical system security frameworks, and software-defined networking environments.