Abstract:

Unfair business competition is more and more often resorting to industrial espionage, which uses more and more new technologies to obtain significant information or technologies. Apart from attacks on IT systems or simple copying of data by bribed employees, there are cases of installation of wiretaps or attempts to intercept data by electromagnetic emission. Protection against conducted or radiated emissions covers mainly the systems significant for the national security, and the systems ensuring this type of protection are designated as TEMPEST class systems. In the case of emission radiated by consumer / industrial devices and wiretaps, a big problem is the detection of this type of emission, the location of the emitter and, in the long term, neutralization / reduction of the emission level. Typically, the analysis of the radio spectrum is performed by a device called a spectrum analyser or a measurement receiver. The problem with using this type of device is that the interpretation of the results, i.e. the detection of signals in a wide spectrum of frequencies, must be performed by a human. The conference paper presents the current, preliminary course of work on the use of machine deep learning to detect radio signals in the radio spectrum, for later use for autonomous observation of the radio spectrum, without the need for interpretation of the results by the operator.