Certain classes of optimization‐based control problems stated for networked systems involving hybrid dynamics and logical constraints can be cast into Mixed‐Integer Programming (MIP) problems. Since these belong to the complexity class NP‐hard, the motivation arises to find approximations of the optimal solution by distributed solution efficiently. For the cases that the cost functional is linear or quadratic and the constraints are linear, this paper proposes an alternative to the standard centralized schemes, by employing dual decomposition into a set of local problems of moderate size which can be solved in parallel. Numerical examples demonstrate that the scheme can efficiently approximate the global solution.