Spectrum underlay technique for secondary multicast network is an effective approach in cognitive radio. In such communication schemes, quality of service constraints are satisfied for the secondary users while interference to the primary users is kept under a threshold. Previous literature consists of suboptimum methods for the solution of the individual problems such as antenna and user selection. In this paper, the joint antenna and user selection problem is considered and solved in an optimum manner. Complete problem is formulated for a discrete beamformer which permits linear constraints and cost function resulting a form suitable for mixed integer 0-1 linear programming. It is shown that antenna selection is very effective to improve the user signal-to-noise ratio while user selection can be employed to increase feasibility. The proposed method performs significantly better compared to suboptimum only-antenna and only-user selection methods in the literature. The problem is further modified to design robust beamformers for any l(p) ball uncertainty region by using a linear approximation. Several simulations are done and it is shown that the proposed approach generates significant improvements in the transmit power and the number of serviced secondary users in a quality of service aware scenario. (C) 2015 Elsevier Inc. All rights reserved.