BPSK and QPSK modulation classification with unknown signal level

Modulation classification is an intermediate step between signal detection and data demodulation. It is attempted for a variety of reasons including reconnaissance, surveillance and other intelligence gathering activities. In practice, the level of a received signal is not known and not constant because of many different reasons, such as fading. In such a case, the classifier needs to estimate the signal level first and then take it into account in the classification algorithm to increase the accuracy of classification. In this paper a modulation classification algorithm with unknown signal level has been developed and tested on binary phase shift keying (BPSK) signal and quadrature phase shift keying (QPSK) signal embedded in additive white Gaussian noise (AWGN). The algorithm applies the generalized likelihood ratio test. The receiver first estimates the unknown signal level using the maximum likelihood method. The estimated signal level is then being used in the likelihood ratio test for classification. Simulations show that the proposed classifier has high accuracy and is superior to the classifier without taking into account the variations in signal level. The classification accuracy increases as the observed symbol number or the variance of the signal level increases.