Optimization problem of broadcast beamforming is nonconvex due to rank condition. Semidefinite relaxation (SDR) method is proposed to solve this problem. However, in certain cases, the performance of SDR becomes unacceptable. In this paper, an equivalent formulation of the original problem without rank condition is found. An iterative algorithm which is shown to converge to a rank one solution is presented. Experimental results show that the proposed algorithm is very effective and gives better results than other existing approaches.