CruxML is delivering real-time solutions at the edge of the network using a fusion of reconfigurable computing, custom inference acceleration and low precision deep neural networks. Only this combination of expertise and technology can solve the types of problems faced by defence, space, natural resources and cyber security, that depend on remote, autonomous and automated sensing and response. We are building IP, systems and tools that can be combined with the domain knowledge of our customers to create novel solutions for the type of problems that are considered ‘too hard’ for today’s tech!
CATJAT micro:
Battery power Personal Electronic Counter Measure device for protection against Improvised Explosive Devices. Utilises advanced Deep Neural Network technology for identification of IED threats by detecting and classifying RF triggering emissions.
CruxSEI:
An edge-based Specific Emitter Identification (or Radio Frequency Fingerprinting) system embedding state of the art machine learning for spoofing detection in NAVWAR applications. CruxSEI can be used for maritime AIS spoofing detection applications in border protection, fisheries, shipping and space domain awareness.
CruxAMC:
Automated Modulation Classification (AMC) plays a pivotal role in non-cooperative communication systems. In the case where a receiver has no prior information about the type of signal AMC can be used to identify the modulation scheme being used. This is crucial for systems like cognitive radio networks, where flexibility and adaptability are key. CruxAMC is an “out-of-box” model for TensorFlow that can be easily integrated in existing ML based systems or added to CPU/GPU or FPGA based systems.
Marsupium:
A Machine Learning data-collection, curation and development framework. Data collected from edge devices are archived on public or private cloud infrastructure for redundancy and firmware updates to edge collection devices can be pushed and managed.
CruxVision:
A Radio Frequency scene understanding capability.