Research On Multi-agent Based Multi-sensor Resource Management

With the rapid development of science and technology, sensor management has become an indispensable part of information fusion system nowadays. In reality, sensors are usually from different platforms, when using multiple sensor platforms for monitoring, coordination between multiple platforms should be considered. In the case of multiple platforms, sensor allocation can be divided into two layers: One is the allocation of combat missions, namely assigning all the sensor platforms to the corresponding combat mission; the other one is for a single combat mission, allocating multi-sensor resources to the target to be tracked.The main works are as follows:(1) The relationship between multi-sensor information fusion and multi-sensor management is discussed first. Then the function of sensor subsystem, information fusion subsystem, human-computer interaction subsystem and sensor management subsystem, which are in the information fusion model of closed-loop control, is discussed.(2) Multi-Agent technology is introduced to the traditional information fusion system. A new information fusion system model based on multi-Agent technology and an information fusion system running mechanism with sensor resource management functions is constructed. Through communicating, cooperating and competing, the resource allocation task between each agent is completed.(3) A resource allocation method based on genetic algorithm is proposed. Through applying genetic algorithm to the allocation of combat platform, establishing the combat mission allocation model, and designing the fitness function, combat mission is assigned to every combat platform. This method solved the allocation problem from the macro, it has laid the foundation for the target distribution in single combat mission.(4) A resource allocation method based on auction algorithm is proposed. Then the general theory and algorithm of auction algorithm is discussed. Compared with the traditional methods of information entropy, the threat level of target, effectiveness function, cost function and value of each sensor platform is considered, the reasonable of allocation between the sensor and the target is improved.The algorithms above are simulated, the simulating result proves that the method is more realistic and effectively meeting the needs of sensor resource allocation.