CAJ | 학술논문

针对ZigBee网络性能评价中多元指标权重确定容易存在的人为性, 以及指标间不可能完全独立的特点, 将熵权法、变异系数法应用于客观权重决策中, 并引入指标间相关性对算法进行改进, 建立基于指标间相关性的多评价指标权重决策分析模型。在仿真实验中, 通过改变节点规模获取网络性能指标数据样本。对比分析表明, 客观赋权方法可以避免易存在的人为性, 引入指标间相关性后, 弱化了异常值带来的影响, 使评价结果更为准确合理。
침대ZigBee망락성능평개중다원지표권중학정용역존재적인위성, 이급지표간불가능완전독립적특점, 장적권법、변이계수법응용우객관권중결책중, 병인입지표간상관성대산법진행개진, 건립기우지표간상관성적다평개지표권중결책분석모형。재방진실험중, 통과개변절점규모획취망락성능지표수거양본。대비분석표명, 객관부권방법가이피면역존재적인위성, 인입지표간상관성후, 약화료이상치대래적영향, 사평개결과경위준학합리。
In ZigBee network performance evaluation, there was artificiality to determine weights of multi-dimensional indexes, and the indexes could not be completely independent. To solve these problems, this paper used entropy weight method and coefficient of variation method, introduced the correlation of the indexes to improve algorithm, built the model of multi-evaluation index weight decision analysis based on the correlation between indexes. Using network simulation software, it collected different network performance parameters by changing the node size. The data analysis show that objective weighting can avoid easily existed artificiality. The introduction of correlation between indicators weakens the impact of abnormal values, which makes the evaluation more accurate and reasonable.