信息估计融合中的观测量化解决方案

在分布式无线传输网络中的估计问题，由于诸多限制条件，传输量十分有限，传感器只能传送有限的几个0-1码（bit）到融合中心，在此情况下完成多传感器估计融合。采用将区间划分，每一区间形成一个假设，规定假设之间合适的损失系数，将估计融合问题化为决策融合问题，运用成熟的决策融合的方法，对于现在研究还不成熟的估计融合传感器量化问题设计了一个解决方案，将决策融合的决策结果作为最终的估计，从而在观测被量化的条件下达到较优的估计效果。并用计算机模拟的例子对分析进行了验证。

The Scheme for Observation Quantization in Information Estimation Fusion

In decentralized wireless sensor networks,the final estimation fusion was accomplished in limited bandwidth channel network,in which each sensor often transmitted a few bits to the fusion center because of various constrains.In order to design a scheme to quantify the observation and to get the fusion estimation from the received bits,the estimation fusion problem was changed into that of decision fusion.The interval was divided into subintervals,each subinterval was assigned a hypothesis,suitable cost coefficients of each hypothesis was selected to form the Bayes cost functional,and a mapping from estimation fusion to decision fusion was constructed.The results of simulation proved the theoretical analysis.