As a consequence of the increasing requirements for more knowledge and information about the actual situations in crisis and war situations new system-architectures will in the near future be needed for the design of sensor-nets. Such systems that include a large number of sensors will with necessity become very complex and achievements for the development of such systems are therefore very important. These systems must be simple to maintain and to expand. Traditional system-architectures do not respond to these demands and for this reason new approaches are needed. In this report, a systemarchitecture with characteristics that respond to the given requirements necessary in modern reconnaissance, control and command systems attached to network of sensors will be presented. Basic elements in the architecture are intelligent agents, which basically can be described as autonomous program objects that independently or in groups can solve different assignments such as distributed sensor data fusion. This report describes the intelligent agent concept and the use of intelligent agents in complex systems. International research activities focusing on intelligent agents in sensor systems will also be discussed. More specifically, the report also describes how these agents can be used in unattended ground sensor-nets (UGS).