Abstract—In this paper a universal demodulator based on probabilistic neural network is presented. It is a kind of modulation (schemes) free demodulator, i.e. it can be trained for different schemes of digital modulation in order to detect incoming data bits without changing or replacing receiver hardware so it can be easily used in software defined radio structure and military communication. Furthermore it has some advantages over the other types of neural network demodulators such as fast training and fast data processing ability to detect data bits since there is no feedback layer in its structure. There is also no need to design special kind of filters except for those which are used to limit the input noise power.
Index Terms—Demodulator, Probabilistic Neural Network, Simulation.
Cite: Mohammad Reza Amini, Einollah Balarastaghi, and Boroujerd Branch, "Universal Neural Network Demodulator for Software Defined Radio," International Journal of Engineering and Technology vol. 3, no. 3, pp. 263-268, 2011.