Blind identification of digital communication signals based on statistics of directional data

This paper presents a new clustering algorithm to solve the blind identification problem of digital communication signal modulation types. the algorithm utilized the instantaneous frequency and instantaneous phase of sampling circular statistical data as a training sample to extract classification feature based on the statistical theory of directional data. And use the characteristic parameters to achieve a variety of different types of communication signals in the modulation recognition on the two-dimensional plane. This method is not only able to make identification classes, but also to achieve recognition subclasses. Simulation results show that the algorithm simple and efficient, and high in accuracy, robustness better and has strong practicality and feasibility.