Speaker
Description
As unmanned aerial vehicles (UAVs) become
increasingly ubiquitous across critical applications, secur-
ing their operations in real time is paramount. This study
presents a formal monitoring framework designed to detect
UAV behavior and anomalies on the fly using Temporal
Streaming Specification Language (TeSSLa). Specifically,
the study focuses on analyzing complex, asynchronous
event and state streams generated by multiple drones
during flight.The system architecture seamlessly combines
an FMI co-simulation engine with an observation attack
layer and an out-of-process TeSSLa monitor for runtime
verification the study presents SWARMRV, a reproducible
harness for runtime security monitoring of swarms under
sensor and telemetry level attacks. It combines an FMI
co-simulation of the swarm, an observation-only attack
layer (GPS spoofing, jamming, and false-data injection)
that falsifies what the swarm observes without touching
the plant, and an out-of-process TESSLA stream monitor
emitting per vehicle/drone and swarm level verdicts over
a four class property taxonomy. Each attack is mapped to
the GNSS/MAVLink signal it abstracts. Over five seeds the
monitor is verdict-quiet on nominal flight (0 false alarms in
3005 steps), detects each attack’s primary signature within
one 50 ms step, adds negligible median overhead with a
bounded tail, and keeps a near flat 95th percentile latency
(30.6–33.1 ms) as the swarm grows from 8 to 32 drones.