A Unified Estimator for Near-Field and Far-Field Localization Using TDOA

讲座名称：A Unified Estimator for Near-Field and Far-Field Localization Using TDOA

讲座时间：2020-01-06 9:30:00

讲座地点：亚博足彩新科技楼1012

讲座人：Dominic K. C. Ho

讲座人介绍：

Dominic K.C.Ho教授现任职于美国密苏里大学电气工程与计算机科学系教授，IEEE Fellow，曾担任IEEE Trans. on Signal Processing、IEEE Signal Processing Letters副主编，IEEE信号处理协会传感器阵列和多通道(SAM)技术委员会主席(2015年)，IEEE ICASSP技术联合主席(2016年)，在信号处理领域有很深的学术造诣并享有广泛的国际名声；是22项关于无线通信、地理位置定位和信号处理领域的美国、加拿大、欧洲专利发明人，发表IEEE TSP/TWC等期刊论文117篇、IEEE ICASSP等会议论文160篇，75项国际电联ITU-T标准或提案的起草人或报告人，美国国家自然科学基金长期资助人员；同时担任我校信号定位与无线通信领域国际合作专家，是亚博足彩客座教授，联合研究方向包括面向人工智能的固态体定位系统、天基频谱感知目标定位系统等。

讲座内容：

Localization of an object is fundamental in many engineering applications. Using of a number of sensors and observing the signal arrival time differences, we are able to determine the unique coordinates of an object if it is not distant. On the other hand, we can only obtain its DOA if it is far away. One has to know if the object is near or far to apply point or DOA estimation and such knowledge is often unavailable. Using a near-field model to locate a distant object will yield an unreasonable result caused by the thresholding effect and applying a far-field model to a near object will produce a significant DOA bias. This talk revisits the HBB Bound that illustrates the thresholding phenomenon in locating the point position of an object as its range to the sensor array increases. A unified TDOA positioning estimator based on an iterative implementation of the MLE is next introduced, which does not require the knowledge if the object is near or far. It automatically generates a point position if the object is near and yields the DOA if it is far. An initialization to the MLE based on a solution by the semidefinite relaxation is proposed. Simulation examples validate the good performance of the proposed estimator.

主办单位：通信工程学院