A large number of control problems e.g. in chemical processing, power distribution, transportation and economic systems can be subdivided into smaller problems, which can be handled by the local controllers in a distributed manner. Each local controller will then be responsible for a subsystem, or agent, which is in fact a part of the whole system. Since the agents are physically interconnected, a communication medium must be existing among the local controllers in order that the individual solutions ultimately reach a consensus, which is optimum from the perspective of the entire system, e.g. in terms of the overall energy consumption or total cost. The advantages of the distributed control approach include respecting the privacy restrictions of the agents, modularity, parallelization of the computations, and improvement in the fault detection and system maintenance. Nowadays one of the widely considered instances of the distributed control is distributed model predictive control (DMPC), which is in fact a networked variant of the well-known model predictive control (MPC) approach. The MPC problems are intrinsically demanding from the computational viewpoint. In DMPC this difficulty emerges as the necessity of transferring huge amounts of data among the local controllers. In this dissertation several DMPC approaches are introduced which aim at decreasing the communication effort required to solve the distributed optimization problem. These novel approaches are examined thoroughly from the feasibility, stability and performance perspectives. Moreover, to investigate the effectiveness of the proposed methods several numerical simulations are carried out using different benchmark systems and the results of the simulations are discussed in depth.weiterlesen