| 摘要: |
| 由于水下传感器网络(Underwater Sensor Network,USN)的能量、带宽有限，传输原始量测数据前需要进行量化处理。面向目标跟踪，在传输比特数据量的约束下，提出了非短视量化比特分配算法。首先，推导了量化量测下的条件后验克拉美罗下界，并将其设为优化目标，建立了比特分配优化模型。在此基础上，提出了一种双层近似动态规划的算法来实现比特分配的优化，在所设时间窗内利用第一层近似动态规划分配各个时刻的比特，并利用第二层近似动态规划在各分支上实现水下传感器节点的比特分配，进一步提升了计算效率。仿真结果表明，所提算法在满足实时性的要求下具有更稳定的跟踪性能。 |
| 关键词: 水下传感器网络(USN) 目标跟踪 非短视比特分配 动态规划 |
| DOI：10.20079/j.issn.1001-893x.220210010 |
|
| 基金项目:国家自然科学基金项目(61871203)；江苏省研究生科研与实践创新计划项目(KYCX21_3501) |
|
| A non-myopic bit allocation algorithm for target tracking in underwater sensor networks |
| TIAN Shengkai,ZHANG Zhenkai |
| (School of Naval Architecture & Ocean Engineering,
Jiangsu University of Science and Technology,Zhenjiang 212100,China;Ocean College,
Jiangsu University of Science and Technology,Zhenjiang 212100,China) |
| Abstract: |
| Due to the energy and bandwidth limitations of underwater sensor networks(USNs),the original measurement needs to be quantized before transmission.For target tracking,under the constraint of total bit,a non-myopic bit allocation algorithm is proposed.First,the conditional posterior Cramer-Rao lower bound(CPCRLB) under quantized measurement is derived and set as the optimization metric.Then a bit allocation optimization model is built.On this basis,a double layer approximate dynamic programming algorithm(DA-DP) is proposed to allocate bit.In set time window,the first layer is used to allocate bit at each moment,and the second layer implements the bit allocation of each underwater sensor nodes on the branch,which further improves the computational efficiency.The simulation results show that the proposed algorithm has better and more stable tracking performance under the requirement of real-time. |
| Key words: underwater sensor network(USN) target tracking non myopic bit allocation dynamic programming |
