Abstract: Constant power algorithm (CPA) is a blind beamforming algorithm used in frequency-hopping spread spectrum (FH-SS) communication systems. However, it suffers two problems. One is that its convergence speed is slow because the target power value needs to be included in the cost function to be optimized as a parameter. Another problem is that it may capture constant power (CP) interference rather than the CP signal of interest (SOI) when such interference as wideband interference or tracking interference in the environment is CP. In order to overcome the shortages, this paper proposes a novel beamforming algorithm, named Differential Constant Power Algorithm with Cycl-correlation Constrains (CCDCPA), which combined two characters of the de-hopped frequency-hopping signals, that is constant power character and cyclostationary character. A reasonable cost function of the array vector is established followed with the deduced iterative formula in this paper. The simulation results show that convergence of the proposed algorithm is faster than that of CPA. Moreover, the proposed algorithm can effectively suppress not only narrowband interference but also wideband interference. Thus, it is an appropriate blind beamforming algorithm used in frequency-hopping system.