通讯作者，Email：zhangghui1230@163.com（衡阳师范学院数学与计算科学系， 中国 衡阳421002）

摘要针对目标搜索传感器调度问题，归纳总结了基于Shannon熵、KullbackLeibler熵和Rényi熵3种信息测度以及全局和局部两种信息增益，衍生出6种基于信息增益的搜索调度模型.通过数值仿真计算对比分析6种基于信息增益的搜索调度模型的目标捕获性能，结果表明基于3种信息测度的搜索调度模型对目标的捕获性能差别很小，而基于全局信息增益的搜索调度模型性能更优，仿真结果可为工程实践中传感器搜索调度方法的优选提供参考.

关键词搜索调度;熵;信息增益;累积捕获概率

中图分类号TN9111文献标识码A文章编号10002537（2015）06007805

Comparison of the Information Gain Based Sensor Search Scheduling Models

ZHANG  Guohui*

（Department of Mathematics and Computational Science， Hengyang Normal  University， Hengyang 421002， China）

AbstractTo deal with the problem of sensor search scheduling， three typical information measures based on Shannon entropy， KullbackLeibler entropy and Rényi entropy are summarized. Additionally， based on both of the  global information gain and local information gain， 6 kinds of search scheduling models are developed. The acquisition performance of all information based scheduling models are compared through simulation. The results indicate that， as for the three typical information measures， the acquisition performance of the search scheduling methods shows the same rules， while search scheduling model based on the global information gain shows better performance. The simulation results can be used as the reference for the search scheduling method in project.

Key wordssearch scheduling; entropy; information gain; cumulative acquisition probability

目标最优搜索理论起源于二战时期美国海军的反潜作战[1]，其动机是增强飞行员对上潜到水面充电的潜艇的检测能力.Koopman[2]对搜索论进行了系统的总结，给出最优搜索问题的3个基本元素：目标位置先验信息、搜索资源与检测概率之间的关系方程及搜索资源约束，并分别从运动学基础、目标检测及搜索资源最优配置角度对搜索理论进行了阐述.学者们针对目标最优搜索理论和相应的传感器搜索调度方法进行了大量研究，涉及雷达[2]、声纳[3]、可见光[4]、红外[5]、射电望远镜等主/被动探测设备，所建立的搜索调度模型包括累积发现概率模型[6]、信息增益模型[2]、自适应搜索模型[7]、分区搜索模型[8]等，这些研究成果已应用于许多不同的领域，如：水下目标或空间目标搜索、地面或海面搜救、地质……