Multi-class classification of analog and digital signals in cognitive radios using Support Vector Machines

Cognitive radio systems require often a capability to recognize used waveforms either for communications or detection purposes. This is a shared problem with many military signal analysis systems. In this paper we present a study of multi-class signal classification based on automatic modulation recognition through Support Vectrom Machines (SVM). We have implement a simulated model of an SVM signal classifier trained to recognize seven distinct modulation schemes; five digital (BPSK, QPSK, GMSK, 16-QAM and 64QAM) and two analog ones (FM and AM). The signals are generated using realistic carrier frequency, sampling frequency and symbol rate values, and pulse-shaping filters types. We also report on our on-going experimental work by using software defined radios (SDR) to implement this as a part of our cognitive radio network testbed. The results show that overall performance of classifier is very good; the trained SVM correctly classifies signals with 85-98% probability in its current state of development.