Method of decomposition in optimization is divided into model coordination and goal coordination. But these two coordinations have problems because of their displacement constraint. This study made an attempt to solve these problems, an attempt which tried out optimization by sorting subsystem problem and system problem from optimum problem and even optimized geometric shape to reduce the weight of truss structure. This study took into consideration to solve system problem only displacement constraint which has a great effect on optimization, and critical constraint of subsystem problem linearized. And to solve subsystem problem it took into consideration allowable stress and buckling stress constraint, which has a little effect on optimum design for other parts of the structure, and constraint of the system problem linearized. Therefore this study which is connected subsystem problem with system problem and at the time governed optimum result of subsystem problem analyzed validity and convergence from four kinds of truss structures. The findings of study are as follows: 1. Dividing optimum problem into system problem and subsystem problem and optimizing these problems, this study reduced non-linear degree. Model and goal coordination hardly took into consideration displacement constraint, but the algorithm applied by this study could easily include displacement constraint into optimization problem. Objective function of system problem is different from that of subsystem problem so that parts of the structure which have different functions could be optimized. independently. 2. Optimum geometric shape is different for the same structure in accordance with constraint. 3. Comparing algorithm of this study with other algorithms, the result of objective function makes little difference and coordinate of joint also makes difference. But optimum geometric shape is similar each other. 4. Optimizing geometric shape, the weight of structure could be considerably reduced. Therefor shape optimization introduced by this study can be used for the economical design.