Research on Blind Demodulation System based on Deep Learning and Software Defined Radio

Blind demodulation systems capable of demodulating signals without modulation information have significant application potentials in many fields such as cognitive radio and electronic warfare. Existing blind demodulation systems usually suffer from demodulation errors resulting from modulation identification errors and bad extensibility due to the limitation of hardware platform. This paper designs and realizes a blind demodulation system based on deep learning and software defined radio. The designed system uses the latest deep learning algorithm instead of traditional feature based algorithms to achieve better accuracy of modulation identification. Moreover, it adopts software defined radio platform instead of traditional circuits to facilitate the upgradation and extension of system functions. Test results show that, the system realized in this paper meets design requirements well, is able to demodulate signals correctly in various scenarios, and can respond to the change of modulation type in time.