Intrusion-and-Impact: Forecasting UAV Communication Attacks and their short-horizon operation degradation

Lakshmikanthaiah, Sahana (2025) Intrusion-and-Impact: Forecasting UAV Communication Attacks and their short-horizon operation degradation. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (701kB) | Preview
Preview	PDF (Configuration Manual) Download (767kB) | Preview

Abstract

The communication links of Unmanned Aerial Vehicle (UAV) are susceptible to cyber intrusion capable of compromising the Quality of Service (QoS). The dissertation introduces a short horizon forecasting pipeline that pipeline which (i) predicts the likelihood of an intrusion occurring within a 3-minute timeframe and (ii) forecasts near-term degradation in the form of quality of service (RTT, jitter, throughput) over a 3 minute time interval. Temporal causality is imposed by strictly chronologically partitioned and leak-safe feature engineering consisting of current-state KPIs, short lags, as well as compact rolling statistics. Both classification and regression are implemented by gradient-boosting models (XGBoost, LightGBM, CatBoost), and the selected classifier is tuned to give the probabilities, which can be trusted. The weighted severity term derived using the absolute KPI deltas, plus the calibrated intrusion likelihood, produces one operational risk score, which puts imminently large impact windows in focus. The classifier is characterized by a discriminative performance on a held-out test segment, and calibration is improved in terms of quality of probabilities (reduced Brier error, and improved reliability curves are closer to the diagonal). The short-horizon regressors also satisfy low absolute error of RTT and throughput values but jitter is more erratic but directional. All in all, the unified risk score is effective to prioritize pre-emptive intervention, and the method is computationally efficient, interpretable and it is consistent with real-time UAV operations.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Simiscuka, Anderson UNSPECIFIED
Uncontrolled Keywords:	UAV communications; intrusion forecasting; QoS degradation; short-horizon prediction; gradient boosting; probability calibration; temporal leakage; risk scoring
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	08 Sep 2026 08:56
Last Modified:	08 Sep 2026 08:56
URI:	https://norma.ncirl.ie/id/eprint/9875

Actions (login required)

View Item