基于?? 均值聚类的二进制传感器网络多目标定位方法

针对存在错误报警的二进制传感器网络,提出基于K均值聚类的二进制传感器网络多目标定位方法。在目标和节点间距离信息未知的条件下,提出基于K均值聚类的改进加正减负算法(KMC-ISNAP)对目标位置进行估计,引入影响因子降低分类过程中模糊节点对多目标定位误差的影响。仿真实验表明,K均值聚类方法在多个目标随机分布情况下能够对报警节点进行准确分类,与质心估计算法和加正减负算法相比, KMC-ISNAP多目标定位方法具有较高的定位精度和较好的容错性。

Multi-objective localization method based on K-means clustering in#br#binary sensor networks

A multi-objective localization algorithm based on K-means clustering is proposed in binary wireless sensornetworks with false alarm. The K-means clustering-improved subtract on negative add on positive(KMC-ISNAP) algorithmis applied to localize the multiple objectives where the distance between nodes and objectives is unknown, and influencingfactors are used to reduce the influence of fuzzy nodes on localization errors. The simulation results show that theK-means clustering method is able to divide the alarmed sensors into parts accurately when multiple objectives arerandomly distributed, and the proposed KMC-ISNAP has higher estimation accuracy and better fault tolerance than centroidestimator(CE) algorithm and subtract on negative add on positive(SANP) algorithm.