测距优化与能量有效覆盖的三维水声传感器网络目标定位跟踪技术研究

For the problem of low precision and high energy consumption in the technology of target localization and tracking in 3D underwater acoustic sensor networks (UASNs), we consider defining the scale factor based on the classification of line-of-sight (LOS) propagation and none-line-of-sight (NLOS) propagation for underwater acoustic signals. And on the basis of optimization, accurate sound travel time solution is designed to improve the ranging accuracy. To detect targets by high probability, the theory of probability and energy effective coverage are introduced. Further to reduce energy consumption of UASNs and optimize the prediction area, grid type measure is proposed. We present the color sequence localization, where we deduce the sailing direction and velocity of the targets based on the model of 3D hierarchical projection. The signal values are stored as the sequence of a target, and by comparing the difference between samples’ sequences and targets’ sequences, samples can be filtered out. The new weighted approach is studied as well to improve the accuracy of sample state estimates. New navigation strategy will be studied for tracking underwater targets dynamically. Meanwhile, based on the combination of software simulation and hardware test bed, the unified and reasonable evaluation system will be studied to verify the tracking probability, tracking error, tracking delay and tracking cost of the proposed algorithms.

针对三维水声传感器网络(UASNs)目标定位跟踪技术中的定位不精和能耗过高问题，本项目拟基于水声信号传播的视线(LOS)与非视线(NLOS)法定义比例因子，并设计精确的基于优化策略的水声传播时延估计方案以提高测距精度。引入概率能量有效覆盖理论，对目标进行概率探测，提出格栅式测量优化预测区域，降低网络能耗。提出颜色序列定位法，在三维分层投影模型下，推导目标的移动方向与速度，同时将信号值存储为目标序列，通过比较样本的颜色序列与目标序列的差值过滤样本点，提出新的加权方法，提高样本状态估计的准确性；并设计新的导航策略，动态跟踪水下目标的运动轨迹。同时研究统一、合理的评价体系，采用软件仿真和硬件平台测试相结合的方式对提出的定位跟踪算法进行跟踪概率、跟踪误差、跟踪延迟和跟踪开销等验证。

在三维水声传感器网络(UASNs)中，目标定位跟踪技术中存在定位不精和能耗过高的问题，本项目在保证原有覆盖能力的前提下，提出了一种新的覆盖增强算法COEA。另外，提出了基于精确声传播时延估计方案的自主水下航行器自定位算法SL-STTS。SL-STTS使用声速的数学期望作为声音的水下传播速度，并在理论上说明了其合理性，利用LMA算法来优化测距从而计算出AUV的位置信息。仿真结果表明，SL-STTS符合克拉美罗界且较对比算法具有更高的定位性能。在原DV-Hop算法上提出了两种新的改进算法来提升DV-Hop的定位精度，分别是HSDV-Hop1和HSDV-Hop2算法。仿真结果表明上述两种方法定位精度比传统算法最大可以提高50%。另外，为了克服基于测距的定位算法对网络硬件要求过高这一难题，同时旨在减小定位误差，本项目在研究中提出了一种基于分级聚类的非测距多边定位法MHC。仿真结果表明，MHC算法在不增加任何通信开销的前提下，定位误差更小，在不同锚节点密度下定位误差减少了25%~50%，运行时间也仅有0.15~0.2秒。此外，本项目还借助加权算法和中值算法，提出了一种改进的跳数调整算法。改进后的算法对传统算法中的每跳平均距离的选取进行了调整，还采用加权法计算节点之间的最小跳数，提高了算法的定位精度。通过仿真证明，该算法在不增加额外的硬件成本和网络洪泛次数的前提下，定位精度比传统算法提高了14.21%。本项目还研究了统一、合理的评价体系，采用软件仿真和硬件平台测试相结合的方式对提出的算法进行误差、延迟等验证。