Sensor Management in a Distributed Environment

Abstract: In this work an investigation of the benefits and problems of implementing a tracker using sensor management is done. The tracker is implemented in a fusion node in a distributed radar simulator provided by Ericsson Microwave. To investigate this, a literature study of sensor fusion and sensor management is first done, after which a practical study is chosen as method. The fusion method presented in this work is then tested so that tests of sensor management, which depend upon implemented sensor fusion, can be trusted. Sensor management is tested by letting the system track a specific target in the simulated environment. The system is tested to see what impact the delay in the distributed environment has on the implemented system’s capability to track an object. Two different scenarios are chosen to test the system, where a scenario in this thesis is a fly-by of two aircrafts in the terrain covered by the radar sensors. To test the actual correctness of the system, three dimensional coordinates of the objects are used and Euclidian distance between the original value and the fused value is used as an error measurement. The results are then displayed in a series of graphs and tables. The results show that the chosen fusion algorithm works well with the unsynchronized data. The delay simulated in the system creates a great uncertainty where the object will be, but the presented prediction algorithm manages to find good estimates of the new positions of the object tracked. Loss of data however forces the system to use less information when estimating positions which leads to loss of track. Even though there is a long time delay the presented prediction algorithm can track the object for a period of time, until it looses track due to loss of data. It is also concluded that a system that manages to track an object using a narrow tracking beam is able to track more objects simultaneously using the same radar sensors.