基于动态贝叶斯网络的WSNs链路质量预测

发布时间：2018-06-27 16:05

本文选题：无线传感器网络 + 链路质量预测　；参考：《工程科学与技术》2017年02期

【摘要】：无线传感器网络中,链路质量预测为数据可靠传输和上层网络协议性能的提高提供支撑。为进一步提高链路质量预测的准确性,提出基于动态贝叶斯网络(dynamic Bayesian networks,DBN)的链路质量预测机制。为避免单一评价指标的片面性,从链路信号质量、链路稳定性及非对称性3方面综合评价链路质量;采用K-means聚类算法对参数进行离散化预处理,得到各参数的离散区间;采用熵值法确定各参数的权重,以消除参数权重计算中主观因素的干扰;为避免最大隶属原则的缺陷,采用非对称贴近度分析法构建综合性的链路质量等级指标;借助贝叶斯网络(Bayesian networks,BN)处理不确定性问题的优势和BN分类器在分类上的良好性能,确定DBN的初始网络和转移网络,采用EM算法进行DBN模型的参数学习,从而构建了基于DBN的链路质量预测模型。实验结果表明了采用非对称贴近度分析法划分链路质量等级的合理性与DBN链路质量预测模型的合理性;与4C及FLI预测模型相比,本文模型具有更高的预测准确度。采用链路信号质量、链路稳定性及非对称性3个指标评价链路质量,采用DBN构建预测模型,可得到更准确及鲁棒性更好的链路质量预测结果。
[Abstract]:In wireless sensor networks, link quality prediction provides support for reliable data transmission and improved performance of upper layer protocols. In order to improve the accuracy of link quality prediction, a link quality prediction mechanism based on dynamic Bayesian networks is proposed. In order to avoid the one-sidedness of single evaluation index, the link quality is evaluated comprehensively from three aspects: link signal quality, link stability and asymmetry, and K-means clustering algorithm is used to discretize the parameters to obtain the discrete interval of each parameter. The entropy method is used to determine the weight of each parameter in order to eliminate the interference of subjective factors in the calculation of parameter weight, and to avoid the defect of maximum membership principle, asymmetric closeness analysis method is used to construct the comprehensive link quality grade index. Based on the advantages of Bayesian Networks (Bayesian Networks) in dealing with uncertain problems and the good performance of BN classifier in classification, the initial and transfer networks of DBN are determined, and the parameters of DBN model are studied by EM algorithm. The link quality prediction model based on DBN is constructed. The experimental results show that it is reasonable to use asymmetric closeness analysis method to divide link quality grade from DBN link quality prediction model, and compared with 4C and FLI prediction model, the proposed model has higher prediction accuracy. Using link signal quality, link stability and asymmetry to evaluate link quality, and using DBN to build prediction model, more accurate and robust link quality prediction results can be obtained.
【作者单位】：南昌航空大学物联网技术研究所;南昌航空大学软件学院;南昌航空大学信息工程学院;
【基金】：国家自然科学基金资助项目(61363015;61262020;61501218;61501217) 江西省高等学校科技落地计划资助项目(KJLD14054) 江西省教育厅科学技术重点资助项目(GJJ150702)
【分类号】：TP212.9;TN929.5

【相似文献】