Under the assumption that the optimal sensor rules can be found for any given fusion rule, we analyze the conditions resulting in the performance equivalence and superiority between fusion rules for general distributed multisensor decision systems. To obtain globally optimal performance of a system, by applying the above results, we can partition all possible fusion rules into equivalence classes of fusion rules and compare performances between some of the equivalence classes; therefore, the number of the valuable fusion rules will be reduced greatly. Moreover,the above analysis does not depend on the statistical properties of the observation data as well as the objective of optimizing a certain system performance.%在任何融合律给定后最优传感器律能求得的假设下,我们分析了导致融合律之间等价性和优越性的条件.应用如上结果,欲获全局最优的系统性能,我们可以划分所有可能的融合律为若干等价类和比较某些等价类之间的性能,于是有价值的融合律等价类数目将大大减少,而且上面的分析并不依赖于观测数据的统计性质和优化系统性能的目标.
展开▼
