Application of a SAW artificial neural network processor to digital modulation recognition
Vyacheslav Kalinin, D. Kavalov · 2002
The architecture of a SAW processor based on artificial neural network is proposed for automatic recognition of different types of digital passband modulation. Three feedforward networks are trained to recognize filtered and unfiltered BPSK and QPSK signals as well as unfiltered 16QAM signals. Performance of the processor in the presence of additive white Gaussian noise (AWGN) is simulated. The influences of second-order effects in SAW devices, phase and amplitude errors on the performance of the processor is studied.