An advanced neural network method of intrusion detection in MANET class mobile radio networks is introduced in the article. It is based on an existing method of intrusion detection in fixed computer networks using self-organizing map. The improvement involves quantity reduction of neural network analyzers and serial-to-parallel structural operation change of neural network analyzers which enabled operation of the method in real-time mode. In order to improve competition capabilities and studying of selforganizing map, counting works of each network neuron potential and network errors have been implemented. Experimental studies have been conducted and level of intrusion detection using the established method have been defined.
mobile radio network, MANET, security mobile radio network, method of detection, Kohonen network