| 摘要: |
| 本文提出了一种基于帕累托多任务学习的到达角（DOA）与到达时间（TOA）联合估计方法，将传统的多任务学习转化为多目标优化问题进行求解.该方法设计了一个轻量化的多任务网络，将多径环境中的DOA与TOA联合估计问题建模为多任务、多标签的回归任务，引入受偏好向量引导的帕累托优化方法，将其进一步分解为一组具有不同权衡偏好的约束子问题，通过并行求解这些子问题，最终能够获得一组具有代表性的帕累托最优解.实验结果表明，与其他多任务学习方法相比，本文所提方法为DOA和TOA联合估计问题提供了一种高精度且灵活的解决方案. |
| 关键词: 到达角（DOA） 到达时间（TOA） 联合估计 多任务学习 多任务网络 帕累托优化 |
| DOI：10.20192/j.cnki.JSHNU（NS）.2025.02.013 |
| 分类号:TP391.4 |
| 基金项目: |
|
| Joint DOA and TOA estimation method based on Pareto multi-task learning |
|
CHEN Erqi, WEI Shuang
|
|
College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 201418, China
|
| Abstract: |
| A joint estimation method for direction of arrival （DOA） and time of arrival （TOA） was proposed based on Pareto multi-task learning， which transformed the traditional multi-task learning problem into a multi-objective optimization problem for solution. A lightweight multi-task network was designed by the proposed method， modeling the joint DOA and TOA estimation problem in multi-path environments as a multi-task， multi-label regression task. Furtherly， a Pareto optimization method guided by preference vector was introduced， decomposing the problem into a set of sub-problems with different trade-off preferences. By solving these sub-problems in parallel， a set of representative Pareto-optimal solutions could be obtained. Experimental results showed that， compared to other multi-task learning methods， the proposed approach provided a higher accuracy and flexible solution for the joint DOA and TOA estimation problem. |
| Key words: direction of arrival （DOA） time of arrival （TOA） joint estimation multi-task learning multi-task network Pareto optimization |
