Automatic classification of QAM signals in fading channel

In this paper, automatic classification algorithm of QAM signals proposed in the past is discussed and improved from a practical view point. The classification algorithm is based on the log-likelihood function which represents the probability density of the QAM signal amplitude under a multipath fading channel. The performance of classification algorithm is evaluated by computer simulations, and the successful classification rates for 16QAM and 64QAM are about 100% for SNR⩾20 dB. However, this algorithm is somewhat cumbersome from the viewpoint of real-time practice because precise calculation of the modified Bessel function used in the algorithm is time-consuming. In this paper, the method to reduce the calculation time by interpolating arbitrary modified Bessel function linearly is also presented