Abstract: The individual identification of radiation sources is an important way to further obtain the inherent information of communication equipment after the modulation style recognition of non-cooperative underwater acoustic communication signals. In order to solve the problem of individual identification of non-cooperative underwater acoustic communication radiation sources in ocean channels, this paper analyzes the performance differences of underwater acoustic communication transmitters, focuses on describing the nonlinearity of the root raised cosine roll-off filter of the transmitter, and proposes an individual identification method using cyclic spectral contour maps to distinguish different radiation sources in response to the lack of prior information and the characteristics of cyclic stationarity in non-cooperative underwater acoustic communication signals. By taking the binary phase shift keying (BPSK) signal as an example, simulation analysis of cyclic spectral features under different roll down factors is conducted, and the research results show the feasibility and effectiveness of the individual recognition method as signal to noise ratio is not less than 3 dB.