: This paper presents an evolutionary algorithm applicable to the task of device adjustment in smart appliances ensembles. The algorithm requires very little environmental knowledge and is therefore complementary to the commonly applied rule based methods, such as ontologies. In contrast to traditional evolutionary algorithms, the new approach avoids any central processing scheme. Instead, the ensemble settings are distributed physically across all devices such that every parameter resides only in the device to which it belongs. This approach enables the correct handling of the dynamic nature of smart appliances ensembles, as will be shown in the course of the paper.