10.7910/DVN/68SXYKBañados, AlexanderAlexanderBañadosUPD-EEEIHarvard Dataverse2026Computer and Information ScienceEngineeringPhysicsOtherADS-Bflight trajectoryaviation cybersecurityanomaly detectionadversarial attacksdata poisoningPhilippine flightsBañados, AlexanderAlexanderBañadosUPD-EEEIBañados, AlexanderAlexanderBañadosUPD-EEEI2026-06-022026-06-0510932135531278application/ziptext/x-markdown2.0CC0 1.0This dataset contains curated ADS-B flight trajectory data used for studying flight-route preprocessing, trajectory modification attacks, poisoning attacks, and machine-learning detection models. The dataset includes raw and processed flight CSV files for five Philippine route groups: Cebu (ceb), Davao (dvo), Iloilo (ilo), Malay/Caticlan (mph), and Puerto Princesa (pps). The dataset is organized into raw route folders, cleaned per-flight datasets, model-ready sequence-window datasets, and generated poison datasets. Each cleaned flight record includes trajectory and motion-related fields such as timestamp, latitude, longitude, altitude, speed, vertical speed, heading, elapsed time, route progress, positional displacement, acceleration, turn rate, and other derived features. The dataset supports four file-level classes: authentic flights, modified-altitude attacks, modified-speed attacks, and modified-position attacks. Poison datasets are also included for evaluating how trained detection models respond to adversarially generated flight trajectories. This dataset was prepared for research on ADS-B trajectory anomaly detection, adversarial attack generation, poisoning evaluation, and route-specific sequence classification.