CivTAK / GPS
Machine Learning‑Tolerant GPS Spoofing
We modify CivTAK to propagate realistic but falsified CoT locations and evaluate one‑class detectors (IF, OCSVM, XGBoost) across inversion, speed, and shift scenarios.
Paper DraftComputer Science • Security • ML
Hello, I’m Rony.
Ph.D. student at Tennessee Tech. I research application‑level GPS spoofing and defenses in CivTAK/ATAK, along with federated learning for CPS security. I enjoy building clean, practical systems and writing about what I learn.
MS
About
I’m a computer science researcher focusing on adversarial threats in mobile tactical systems. Previously at Lamar University; now at Tennessee Tech under Dr. Amr Hilal. My recent work demonstrates that ML detectors struggle with realistic spoofed GPS signals and explores cryptographic + multi-sensor mitigations.
Interests: GPS/CPS security, Federated Learning, Applied Cryptography, Software Engineering for ML
Research
Federated Learning
Zero‑Trust FL for CPS
Isolation‑Forest–based trust filtering in Flower; compare against FedAvg on Smart Grid Stability data. Logs + global model saved per round.
Project NotesNLP / SE
seBERT & Requirement Classification
Pretrained on software corpora; evaluates zero‑/few‑shot classification for sentiment and requirement types.
Code & ResultsProjects
TAK Server on OCI
Dockerized TAK Server with SSL and device cert management; cross‑network CoT relay.
GPS Spoofing Toolkit
Android hooks for location mutation (shift, inversion, speed). Data collection + visualization.
ZTFL Simulator
Flower‑based FL with trust scores and anomaly logs; saves per‑client + global models.
Selected Publications
Latest Posts
Contact
Email: shafikrony@gmail.com · GitHub: @shafikrony · Google Scholar / LinkedIn links here.