分布式检测融合可以提高检测网络节点的整体检测性能,然而简单地采用某种判决融合方式并不能保证得到最佳的判决结果,必须结合具体的单个节点的检测概率和虚警概率以及节点数量采取相应的融合规则.通过对各节点在相同的检测概率或虚警概率的情况进行分析,得出分布式检测融合判决规则的性质和适用的条件.最后通过具体的实例表明,利用这些性质选择合适的融合规则能够实现提高检测概率、降低虚警概率的目的.
Distributed detection can increase the performance of the whole sensor network, however, it can not ensure to achieve optimum performance by only taking one specific fusion rule in any situations. The detection probability and false alarm probability of single node as well as the node number should also be considered for determining the fusion rule. An analysis of fusion rule for distributed detection is introduced under the assumption that all the sensors are similar and operate at the same level of false alarm probability and detection probability. These properties and conditions can be used to increase the detection probability and to lower the false alarm probability. Finally a computation example is given for further justification.