A Modulation Classification Algorithm for Multipath Signals Based on Cepstrum

Abstract

Automatic modulation classification (AMC) is an important intermediate step between signal detection and demodulation by the receiving instrument in the noncooperative communication system. Multipath fading causes serious degradation in classification performance of AMC. For the purpose of improving the performance of AMC for multipath modulation signals, we proposed a novel cepstrum-based preprocessing algorithm to eliminate the effect of multipath channel coefficient, and a logarithmic functional fitting method to classify modulated signals. First, we mathematically proved the feasibility of the proposed approach. Then, we experimentally tested our algorithm by the practically measured modulation data of quadrature phase-shift keying (QPSK), 16-ary quadrature amplitude modulation (16QAM), and 64-ary quadrature amplitude modulation (64QAM). The results confirmed that our algorithm performed well for multipath signals and was robust to slight frequency and timing offsets. Compared with the existing methods, our algorithm also performed better than the techniques based on high-order cumulants or cyclic statistics and suffered less computational complexity than the cumulant-based method with channel estimation procedure.