多参数未知下水声传感网由粗到精的定位方法  

Coarse-to-fine localization method for UASNs under unknown multi-marameters

作　　者：鲜江峰 马俊领 吴华锋[2] 梅骁峻 谭拂晓 张媛媛 陈信强 王维军 XIAN Jiang-feng;MA Jun-ling;WU Hua-feng;MEI Xiao-jun;TAN Fu-xiao;ZHANG Yuan-yuan;CHEN Xin-qiang;WANG Wei-jun(Institute of Logistics Science and Engineering,Shanghai Maritime University,Shanghai 201306,China;Merchant Marine College,Shanghai Maritime University,Shanghai 201306,China;Logistics Engineering College,Shanghai Maritime University,Shanghai 201306,China;School of Computer Science and Information Engineering,Changzhou Institute of Technology,Changzhou 213032,China;Navigation College,Jimei University,Xiamen 361021,China)

机构地区：[1]上海海事大学物流科学与工程研究院,上海201306 [2]上海海事大学商船学院,上海201306 [3]上海海事大学物流工程学院,上海201306 [4]常州工学院计算机信息工程学院,江苏常州213032 [5]集美大学航海学院,福建厦门361021

出　　处：《控制与决策》2025年第1期38-47,共10页Control and Decision

基　　金：国家自然科学基金项目(52201403,52331012,52201401,52472347,52071200,52102397);上海市晨光计划项目(23CGA61);国家重点研发计划子课题(2021YFC2801002);上海市科委项目(23010502000)。

摘　　要：水声传感网(underwater acoustic sensor networks,UASNs)是水下物联网的主要技术,为海洋生态环境监测和水下搜救等应用提供了较好的技术手段和信息感知平台.在UASNs应用中定位至关重要,因为没有精确位置信息的数据收集将无利用价值.然而,由于存在路径损耗、吸收损耗、设备发射功率不确定以及水下环境参数未知等不利因素,使得在复杂动态海洋环境中实现鲁棒精确定位较为困难.为此,提出一种多参数未知下水声传感网由粗到精的定位方法(coarse-to-fine localization method for UASNs under unknown multi-parameters,CFL-UMP).首先,利用一阶泰勒级数展开和若干近似操作,将原非线性非凸定位问题转化为交替非负约束最小二乘框架;然后,粗定位阶段基于Golub-Kahan双对角化的最小二乘残差(LSMR)算法求解得到位置估计近似解,然而LSMR通常只能快速收敛到局部最优解,因此在精细定位阶段引入二分法,将第1步粗估计得到的近似解作为二分法的初始值,通过迭代同时得到水下目标位置、路径损耗因子以及发射功率的精确解;接着,为了验证CFL-UMP算法的优越性,分析CFL-UMP算法的计算复杂度,并推导出克拉默-拉奥下界;最后,与所选基准算法相比,通过仿真结果证实了CFL-UMP在不同水下模拟场景中均能获得最优的定位精度,且能够有效降低水下定位误差.Underwater acoustic sensor networks(UASNs)are the main technology of the underwater Internet of Things(IoT),providing a better technical means and information sensing platform for applications such as marine ecological environment monitoring and underwater search and rescue.In the application of UASNs,localization is crucial because data collection without accurate location information will be of no use.However,the presence of unfavorable factors,such as path loss,absorption loss,uncertainty in device transmit power,and unknown parameters of the underwater environment,renders it more challenging to achieve robust and precise localization in complex dynamic ocean environment.Therefore,this study proposes a coarse-to-fine localization method for UASNs under unknown multi-parameters(CFL-UMP).First,the original nonlinear and nonconvex localization problem is transformed into an alternating nonnegative constrained least squares framework(ANCLS)using a Taylor first-order expansion and several approximation operations.Subsequently,in the coarse localization stage,the approximate solution is obtained based on the Golub-Kahan bi-diagonalized least squares minimal residual(LSMR)algorithm.However,LSMR typically only converges rapidly to a locally optimal solution.Consequently,the dichotomy method is employed in the fine localization stage.The approximate solution derived from the coarse estimation in the preceding step serves as the initial value for the dichotomy method,and the exact solutions for the underwater target location,the path loss factor,and the transmit power are simultaneously obtained through iterations.Furthermore,to demonstrate the superiority of the CFL-UMP mehod,the computational complexity of the CFL-UMP method is analyzed and the Cramér-Rao low bound(CRLB)is derived.Finally,compared with the selected benchmark algorithms,the simulation results verify that the CFL-UMP achieves optimal localization accuracy in different underwater simulation scenarios,effectively reducing the underwater localization erro

关 键 词：水声传感网 目标定位 接收信号强度 多参数未知 最小二乘残差算法 二分法 

分 类 号：TB393[一般工业技术—材料科学与工程]