There are usually coverage holes caused by uneven deployment of the nodes in wireless sensor network, which brings about many difficulties in various applicationsi as data value in coverage holes can't be detected directly. To solve this problem, the major idea so far is to place new nodes into the coverage holes with respect to the coverage percentage. However, it is hard to cover the whole region by deploying new nodes, when the network size is huge. Recently, a few methods are designed to estimate the data value in the coverage hole using sensed data of neighboring nodes. Motivated by this idea, a new moving-neighborhood interpolation algorithm is proposed in this paper. The algorithm based on Delaunay triangulation technique can find out a neighbor set of points by iterative searching, which has strong spatial correlation with interpolation point. Then, the value of interpolation point is able to estimate by our proposed algorithm, according to the observations of neighbor set. Moreover, an adaptive selection technique is designed for searching different neighbor set to meet different error thresholds. Experimental results on a real-world dataset show that our proposed algorithm can estimate the value of coverage holes more accurately and robustly. Besides, it can help retrieve the real-time value of a point in the sensor network.%由于节点部署不均匀,传感器网络存在覆盖漏洞即监测盲区,监测盲区影响被监测区域环境数据收集的精确性,给无线传感器网络各种应用带来了巨大困难.以往解决这一问题的主要手段之一是从覆盖率出发,部署新节点缩小监测盲区.但随着传感器网络的规模增大,这种依赖部署新节点达到全面覆盖监测环境的方法很难实现.另一种方法则是通过盲区邻域节点对盲区插值预测,评价盲区的数据.为此,提出一种新的滑动邻域插值盲区评价算法,该算法基于三角剖分技术,迭代筛选出与预测点空间相关性强的传感节点集,然后利用节点采样值对预测点进行估算,以获取插值点误差最小的估计值.实验结果表明,所提出的滑动邻域插值算法能有效估计监测盲区预测点的数据与缺失数据,具有可靠、稳定的估计性能,并可支持网内任意点的实时估计.
展开▼
