Abstract: The existing eavesdropping method has poor anti-noise performance and high complexity, which makes it not practical. It is shown that the received scrambled signals are distributed within parallel hyperplanes if adequate antennas are equipped by eavesdropper. According to this distribution, a hyperplane clustering (HC) algorithm was presented to blindly estimate the hyperplane parameters which reveal the sending information. Simulation results show that the HC algorithm, compared with the existing MUSIC-like algorithms, holds the advantages of better anti-noise performance and lower computing complexity.