• distributed parallel computing
• adaptiveness
• distributed optimization

With this thesis we take a step further on improving reconfiguration decisions in adaptive distributed parallel computations. The concept of switching cost is introduced with the aim of reducing the amount of reconfigurations and of improving the reconfigurations stability in dynamic execution scenar- ios. Computation modules control is based on the Model-Based Predictive Control (MPC) approach. We study the effectiveness of this approach in parallel distributed computations, where each module cooperates to find global optimal reconfiguration trajectory. Experimental results are obtained by means of experiments performed in a simulation environment.