In the process of System-of-Systems, the coordinated kill net composed of platforms has a strong adaptive ability to dynamically generated threats and changing environments, which provides the basic support for cross-domain cooperative operations. The modern battlefield environment is full of uncertainty, so it is very important for military command decision to build kill net based on operational demand and current situation of battlefield. In this paper, an optimization algorithm is proposed based on discrete particle swarm optimization, aiming at the problem of kill net generation under the constraint of the topological relationship of the combat network, which can pre-plan before the battle. The population renewal method based on cross variation is constructed in discrete domain, and the self-organizing inertia weight and acceleration coefficient are integrated. All these methods improve the efficiency of both global search and local search, and overcomes the problem of particle swarm algorithm easily falling into local optimal solutions.