Proceedings Article | 20 April 2017
Krzysztof Kulpa, Michał Szczepankiewicz, Marcin Żywek, Mateusz Malanowski, Jacek Misiurewicz, Piotr Samczyński
KEYWORDS: Warfare, Cognition, Radar, Signal detection, Databases, Neural networks, Defense and security, Machine learning, Reconnaissance, Automatic control
Electronic warfare (EW) techniques were invented 70 years ago and are still being developed, all the time with a key role played by their operators. However, a human operator responds far too slowly for the on-going needs since to be adequate to current threats the EW system reaction should be within 1 ms or less. Too slow reaction defers an achievement of the objectives, and generally may lead to disclosure of the applied strategy and to waste of resources. Developing the adequate response to the threat is a multifaceted matter that requires considerable experience and knowledge on the one hand, and thorough observation of outcomes of the applied strategy on the other hand. Such action requires the use of not only intelligence but also more complex algorithms for automatic control than the classical ones, and it heavily builds on the experience. It is therefore a cognitive task in its nature, and a human operator acts naturally this way. Consequently, a proper training and experience gained are what really matters. As a result of tests and actions in EW the operator builds his own knowledge base, thus, gains a capability of responding to the known threats in a balanced way. New threats, however, are much more challenging since they should be handled as fast as possible but in a fresh and creative manner. To this end, adapting the algorithms of radar jamming protection for the new conditions of the battlefield is an example of such challenging tasks. At present, such adjustments are done in a long feedback loop when the operator, after recognizing the problem, notifies the equipment manufacturers, and then, their engineers work on required solutions accordingly. Thus, the reaction time is counted in years, not a single milliseconds or seconds. Speeding up the response time is therefore the key issue, and a solution to it would be feasible only when cognitive systems were used. A concept of such cognitive system is presented in this paper.
