CELLHAWK: A Triply-Redundant Navigation Architecture for GPS-Denied and Electronically Contested Environments

This paper presents CELLHAWK, a software-defined, hardware-agnostic autonomous navigation frameworkdesigned to maintain full mission continuity for Unmanned Aerial Vehicles (UAVs) operating in Global NavigationSatellite System (GNSS)-denied and electronically contested environments. Existing commercial autonomousplatforms fail predictably when subjected to Radio Frequency (RF) jamming or GPS spoofing, as they rely on asingle primary navigation source. CELLHAWK addresses this single point of failure through a triply-redundant,sensor-fused navigation stack integrating: (1) Primary GNSS, (2) passive Received Signal Strength Indicator(RSSI)-based multilateration using ambient 4G/LTE cellular infrastructure, and (3) Visual SimultaneousLocalization and Mapping (SLAM) optical odometry. Transitions between tiers are governed by an ExtendedKalman Filter (EKF) that dynamically scales its innovation covariance matrix based on real-timeJamming-to-Signal Ratio (JSR) measurements at the Software Defined Radio (SDR) front-end. All resultsreported are simulation-derived over real-world 3D terrain at global locations, demonstrating sub-50 m RMSpositioning error maintained across the full jamming spectrum from 0 dB to 30+ dB JSR. The architecture isadditionally validated against real-world UAV combat loss scenarios sourced from open-source conflict reporting,providing adversarially grounded training data for the CORTEX v2.0 AI autonomy engine. Physical SDR fieldvalidation is planned for Q3 2026.