Classification of BPSK and QPSK signals in fading environment using the ICA technique

Automatic modulation classification through signal processing has found many applications in wireless communication systems, including interference identification, spectrum management and electronic warfare. In practical wireless communication, due to the multipath, the level of the received signal is always time-varying and not known. This paper studies the use of independent component analysis (ICA) technique in identification of BPSK signal and QPSK signal without a priori knowledge of the received signal level. The receiver first estimates the unknown signal level using the ICA technique. The estimated signal level is then being used in the likelihood ratio test for classification. Simulations show that the proposed classifier has a high accuracy with small number of observation symbols. The performance is close to the best classifier that assumes the received signal level is known to the receiver.