In this paper, we propose an efficient adaptive beamforming algorithm based on the simultaneous projection in dual domains. In the (primary) domain of beamformer weights, the proposed algorithm operates the projections onto multiple closed convex sets each of which bounds the magnitude of an instantaneous beamformer-output. At the same time, it also operates the projection onto a closed ball in a secondary domain. All the projections are combined convexly in the primary domain. The idea of the secondary-domain projection is to bound the energy of interfering signals remaining in the beamformer output. The numerical examples show that the proposed algorithm achieves drastic improvements of the convergence speed.