Automatic Modulation Classification Based on Kernel Density Estimation

In this paper, we propose an efficient automatic modulation classification (AMC) scheme for a group of narrowband and digitally modulated signals such as quadrature phase-shift keying (QPSK), 16-PSK, 64-PSK, 4-quadratic-amplitude modulation (QAM), 16-QAM, and 64-QAM. The classification was performed by analyzing the probability density distribution for the real and imaginary parts of the modulated signals. To simplify the complexity of the proposed approach, we performed the classification in two stages: first, we classified the modulation between QAM and PSK signaling, and then, we determined the M-ary order of the modulation by developing kernel density estimation, which is typically used in nonparametric methods for the estimation of the probability density function of a random variable with finite data samples. Simulations were carried out to evaluate the performance of the proposed scheme for flat channels. It is observed that this simple efficient technique can find applications in blind AMC, as the performance comparison with the state of the art is promising.