Dynamic cooperative coevolutionary sensor deployment via localized fitness evaluation

Abstract

We propose an innovative cooperative co-evolutionary computation framework, Dynamic Cooperative Coevolution (DCC), which provides dynamic coupling of neighboring species for the fitness evaluation of individuals. One feature of DCC is the utilization of local fitness to achieve a global optimum, which makes it possible for co-evolutionary algorithms to be applied in localized distributed environments, such as network computing. This work is motivated by our interest in autonomous sensor deployment, where a sensor can only communicate with those within a limited range. Our experiments show that DCC is effective in obtaining good solutions under such distributed and localized conditions. © 2008 Springer-Verlag Berlin Heidelberg.