Cognitive EW replaces the fixed library with systems that sense, learn, decide and act in real time. The loop resembles the OODA loop:
- Sense — an unknown emitter is picked up and characterised (frequency, pulse pattern, modulation).
- Learn — the system recognises the type of threat, or establishes that it is something new, and builds a model on the spot.
- Decide — a suitable countermeasure is chosen or even synthesised (a jamming waveform that was not yet in any library).
- Act — the countermeasure is applied and its effect is fed back to adjust the response.
The engine underneath is machine learning: classification of signals, anomaly detection for unknown emitters, and reinforcement learning to optimise the jamming based on the observed effect. In effect, the system “experiments” on the opponent and learns what works — faster than a human or a static library can.
Adaptive EW adjusts its parameters within pre-defined limits (e.g. automatically choosing a jamming profile). Cognitive EW goes further: it can come up with new solutions for situations the designer did not foresee. The difference is one of degree, but the direction is clear: from pre-programmed to learning.