Intelligent video surveillance (IVS) systems are based on so called smart cameras that combine video sensing, processing and communication within a single embedded device. Maximizing the service-level and minimizing the power consumption are two important but conflicting objectives in IVS. In this paper we present a novel software framework for power-aware reconfiguration in distributed intelligent video surveillance (IVS) systems. The multilayer heterogenous software framework hosts services for power-aware system-level task distribution and is based on a publisher-subscriber middleware approach. An online multi-criterion optimizer permanently computes optimal camera configurations with respect to a given cost model for both power consumption and service-levels.