基于感知源的数据驱动信任评测模型

A data-driven trust evaluation model based on the perception source

中文关键词  物联网  信任评测  数据融合  无线传感器网络

英文关键词  internet of things  trust evaluation  data fusion  wireless sensor networks

中文摘要

      为解决物联网数据源头的可靠问题, 构建一种基于感知源的数据驱动信任评测模型. 模型以监测模块为评 测单元, 由中继节点完成其所在监测模块内感知节点的信任评测, 通过感知节点自身数据之间的关系实现直接信任 的计算, 利用监测模块内各邻居节点之间关系实现推荐信任的计算, 再结合历史信任, 输出感知节点的综合信任. 同时与模型预设的可疑阈值和异常阈值进行对比, 更新历史信任和信任列表, 实现感知节点的异常检测, 利用预警 检测误差和失信检测误差对模型的检测效果进行评价, 统计结果表明模型能够保持较低的平均误差. 将信任机制 引入到数据融合过程, 用综合信任作为加权因子, 从而提高了数据融合的准确度. 最后, 通过实验仿真对信任评测模 型进行评价, 结果表明引入信任评测模型后延长了节点开始死亡的时间, 随着节点的更新迭代, 失信节点越来越少, 在一定程度上提高了节点的存活率, 延长了网络的生命周期.

英文摘要

      In order to solve the problem of reliability of source data in IoT, a data-driven trust evaluation model based on the perceptual source is established. In the model, the monitoring module is used as the evaluation unit, the relay nodes completed the trust evaluation of the sensor nodes in the monitoring module, the direct trust calculation is realized by the relationship with the sensor nodes’ data, the recommendation trust calculation is realized by the relationship in the neighbor nodes in the monitoring module, and then combined with the historical trust to output the comprehensive trust of sensor nodes. Meanwhile, compared with the suspected threshold and the abnormal threshold preset by the model, updated the historical trust and trust list to realize the abnormal detection of the sensor nodes. Using the alert detection error and the dishonest detection error to evaluate the detection effect on the model, the statistical results show that the model can maintain a low average error. The trust mechanism is introduced into the data fusion process; the comprehensive trust was used as the weighting factor, thus, the accuracy of the data fusion is improved. Finally, the trust evaluation model is evaluated by experiment simulation, the results show that it prolongs the time of node death after leading into the trust evaluation model. With the update iteration of nodes, the dishonest nodes will be fewer and fewer. It improves the nodes’ survival rate and prolongs the life cycle of the networks to a certain extent.