We consider the downlink of a cellular system in which each base station (BS) has multiple distributed antenna ports that are geographically dispersed over the cell. The goal of the BSs is to improve cell-edge performance by selecting the subset of ports that maximizes the minimum signal-to-interference-plus-noise ratio of the user terminals in a coordinated manner. This problem is cast as a binary-constrained optimization problem, which is known to be NP-hard. To circumvent this difficulty, the semidefinite relaxation technique is used to efficiently generate Gaussian distributed vectors. Simulation results show that rounding a relatively small number of these vectors yields close-to-optimal solutions of the original problem.