Maximum likelihood BPSK and QPSK classifier in fading environment using the EM algorithm

Modulation classification is to identify the modulation type of a received signal automatically. It is a promising technique that can increase the throughput of adaptive modulation systems such as IEEE 802.11 and HIPERLAN systems. While a variety of approaches have been proposed for modulation classification, most of them rely on the unrealistic assumption that the channel is additive white Gaussian noise channel. This paper presents an algorithm using hybrid likelihood ratio test for BPSK and QPSK classification in multipath fading environment. The unknown channel coefficients are first estimated blindly using the EM algorithm. The estimates are then used in likelihood ratio test for classification. Simulations show that the proposed classifier has high accuracy and is superior to the classifier that does not take the channel distortion into account. An added advantage of the proposed algorithm is that the maximum likelihood estimate of the channel coefficients can be used in equalization and data demodulation.