Proposed in this paper is a blind identification algorithm of orthogonal frequency division multiplexing ( OFDM) signals based on multifractal features,which is applied to the modulation identification in non-cooperative communication. In this algorithm,first,two items,namely the singularity exponent of the maximum multifractal value from communication signals and the span of multifractal from the amplitude of twice Fourier transform of
squared analytic communication signals,are extracted as signal feature parameters. Then,a multi-layer decision tree separator is constructed to identify OFDM signals and single-carrier signals. The feature parameters extracted by this algorithm are robust to the number of subcarriers,the length of the cyclic prefix of OFDM signals,the raised-cosine shaping roll-off factor of single-carrier signals and the signal-to-noise ratio ( SNR) ,and the signals are processed in intermediate frequency channel. Thus,the synchronization in time and frequency domains is not required. Simulated results show that the proposed algorithm can identify OFDM signals with a high average identification rate in a wide SNR range.