Content for the HPC community and innovation enthusiasts: tutorials, news and press releases for users, engineers and administrators
- All News
- Aerospace & Defence
- Artificial Intelligence
- Cloud Platform
- E4 Various
- European Projects
- Kubernetes Cluster
- Latest news
Cognitive Signal Classifier: improving the RF spectrum awareness with artificial intelligence
Improving the RF spectrum awareness is critical for Electronic Signal Intelligence (ELINT) and in general Electronic Warfare (EW) applications.
But overall it is the basis of a more efficient spectrum sharing. This is important as more radios, communications systems, radars and many other applications, including internet-of-things devices, operate in the spectrum and EW is only a subset of it.
In this crowded scenario hostile emitters are becoming much more clever at camouflaging their signatures to look like neutral emitters.
Artificial Intelligence systems can outperform traditional approaches in signal classification at every signal-to-noise ratio. We refer to this technology as a Cognitive Signal Classifier.
There are indeed large investments of startups and companies into this specific field, so let’s see a couple of examples that are available on the market. Startups and companies are investing significant resources in this field. Among these we point out the DeepSig (www.deepsig.ai ) company founded in 2016 in Virginia (USA).
DeepSig has developed the OmniSIG™ Sensor solution which uses artificial intelligence (AI) for signal identification and classification and is very effective for electronic warfare and electronic signal intelligence applications.
OmniSIG™ Sensor performs detection and classification of RF emissions over very wide bandwidths of the spectrum. It also provides the ability to report anomalies, changes or threats in real time. It works with both broadband and narrowband signals and delivers accurate results for highly dynamic signals and within contested tactical scenarios.
Detection and recognition work on a wide range of signal types, with which the system has already been trained. But the system can be extended to include additional signals and protocols. This is done through the OmniSIG™ Studio tool which allows you to incorporate custom data, signals and signatures to complete the training of the artificial intelligence algorithms and optimize the classifier according to specific knowledge and scenarios.
OmniSIG™ Sensor can be implemented in various computing architectures and uses graphics processing unit (GPU) acceleration. The choice of architecture affects the dynamics of the classification system and therefore the real-time or near real-time operation. Different tactical and operational scenarios may require different compute architectures, from rack-mount multi-GPU (Nvidia) servers, to Nvidia Jetson SOM-based rugged edge devices.
E4 Computer Engineering has signed a partnership with DeepSig for the European market, and is able to supply the best processing architecture sized for the requirements of the tactical application scenario.