Dealing with an excitation matching method, this paper presents a global optimization strategy for the optimal clustering in sum-difference compromise linear arrays. Starting from a combinatorial formulation of the problem at hand, the proposed technique is aimed at determining the subarray configuration expressed as the optimal path inside a directed acyclic graph structure modelling the solution space. Towards this end, an ant colony metaheuristic is used to benefit of its hill-climbing properties in dealing with the non-convexity of the sub-arraying as well as in managing graph searches. A selected set of numerical experiments are reported to assess the efficiency and current limitations of the ant-based strategy also in comparison with previous local combinatorial search methods.