Advances in embedded systems and mobile communication have led to the emergence of smaller, cheaper, and more intelligent sensing units. As of today, these devices have been used in many sensor network applications focused at monitoring environmental parameters in areas with relative large geographical extent. However, in many of these applications, management is often centralized and hierarchical. This approach imposes some major challenges in the context of large-scale and highly distributed sensor networks. In this paper, we present a multilayered, middleware platform for sensor networks offering transparent data aggregation, control, and management mechanisms to the application developer. Furthermore, we propose the use of multiagent systems (MASs) to create a computing environment capable of managing and optimizing tasks autonomously. In order to ensure the scalability of the distributed data fusion, we propose a three-step procedure to balance the workload among machines using mobile agent technology.