A European Space Agency (ESA) NAVISP-funded project, led by Huld (formerly Space Systems Czech) has developed an advanced RF2RF (radio frequency-to-radio-frequency) GNSS receiver enhancement device as a powerful countermeasure to jamming and spoofing, with real-time detection, classification, and mitigation capabilities.
The Huld Block-box enhances any commercial off-the-shelf (COTS) GNSS receiver by incorporating three critical functionalities. First, it utilizes an onboard artificial intelligence (AI) engine to detect and classify GNSS threats in real time, enabling real-time identification of anomalies with high accuracy. Second, the device employs digital signal processing (DSP) to clean up and retransmit disrupted signals, ensuring seamless operation of both stationary and mobile GNSS assets. Third, the Block-box connects with a cloud-based processing platform, allowing large-scale threat monitoring and AI model refinement when multiple units are deployed.
ESA recently hosted a special event at which project partners from Huld explained their approach, which began with a comprehensive review of existing GNSS threats and countermeasures. New software was developed, implemented in Python, incorporating AI-based threat classification and DSP-based signal recovery. New hardware was based on a field-programmable gate array (FPGA) design. To test performance and reliability, the team conducted extensive trials using synthetic GNSS signals, followed by real-world testing with GNSS recordings from ESA.
New capabilities
The Block-box prototype is equipped with a robust RF signal retransmission system, covering a frequency range from 0 to 3 GHz. It supports up to two independent RF inputs and outputs, with four processing channels capable of handling bandwidths of up to 150 MHz each. Its AI-based jamming detection and mitigation capabilities utilize ResNet and U-net convolutional neural networks (CNNs) to identify interference patterns. The device effectively counters continuous wave, chirp, and pulsed jammers using a frequency-domain adaptive filtering (FDAF)-based DSP method. Additionally, it offers real-time spoofing detection for Galileo E1 and GPS L1 C/A signals, employing a simple ResNet CNN to analyze cross-ambiguity function snapshots.
The Block-box also features data storage and replay capabilities, allowing simultaneous four-channel recording with bandwidths of up to 46 MHz per channel. Project researchers said further real-world data collection is necessary to improve AI model training and performance evaluation. The device’s ability to function in complex environments with multipath effects and signal fading also requires additional validation. Future versions of the Block-box may incorporate advanced antenna configurations, such as phased arrays or dual-polarization antennas, and efforts are underway to design a smaller, more commercially attractive version of the device.
