Project agency (school, industry)
项目机构（学校、行业）

School of Computer Science
计算机科学学院

Industry supervisor and contact emails
行业主管和联系人电子邮件

TO be confirmed
待确认

Academic Supervisor name(s) and contact emails
学术导师姓名和联系邮箱

Wenzong Gao, w9.gao@qut.edu.au

Yanming Feng, y.feng@qut.edu.au
冯艳明，y.feng@qut.edu.au

Information Technology major(s)
信息技术专业

Data Science and Computer Science
数据科学与计算机科学

Project title
项目名称

Precise modelling and prediction for GNSS code multipath delays using machine learning approaches
利用机器学习全球导航卫星系统代码多路径延迟的精确建模和预测方法

Brief description of the research problem, gaps, aims, methodology and expected outputs (~200 words)
简要说明研究问题、差距、目的、方法和预期成果（~200 字）

The Global Navigation Satellite Systems (GNSS), particularly the US's Global Positioning System (GPS), have become integral to daily life, industry, military operations, and scientific research. GNSS technology allows for the precise determination of any receiver's 3D coordinates on Earth's surface, provided the receiver has line-of-sight (LOS) measurements, i.e., receiver-to-satellite distances, to at least four satellites. Using the known positions of these satellites and the measured receiver-to-satellite distances, the receiver’s position can be accurately calculated.
全球导航卫星系统（GNSS），特别是美国的全球定位系统（GPS），已成为日常生活、工业、军事行动和科学研究不可或缺的一部分。全球导航卫星系统技术可以精确测定任何接收器在地球表面的三维坐标，前提是接收器可以测量到至少四颗卫星的视线（LOS），即接收器到卫星的距离。利用这些卫星的已知位置和测得的接收机到卫星的距离，可以精确计算出接收机的位置。

The measurement of the receiver-to-satellite distance depends on the time difference between the signal transmitted by the satellite and received by the receiver. However, various factors can introduce errors into these measurements. For instance, atmospheric delays occur when signals are distorted as they pass through the atmosphere. Additionally, multipath delays, which happen when satellite signals are reflected and diffracted by tall buildings and skyscrapers, often reduce accuracy by several tens of meters. While most errors can be corrected using mathematical or empirical models, multipath errors are challenging to model through traditional methods, making them the most significant source of GNSS measurement errors.
接收器到卫星距离的测量取决于卫星发射信号与接收器接收信号之间的时间差。然而，各种因素都会给这些测量带来误差。例如，当信号穿过大气层时，会产生大气延迟distorted 。此外，卫星信号被高大建筑物和摩天大楼反射和衍射时产生的多径延迟通常会使精度降低几十米。虽然大多数误差可以通过数学模型或经验模型进行修正，但多路径误差却很难通过传统方法进行建模，因此成为 GNSS 测量误差的最主要来源。

This project aims to apply Machine Learning (ML) techniques to model code multipath delays in GNSS code measurements. These delays will be calculated from dual-frequency GNSS measurements, and variables related to the multipath delays will be selected as inputs for training an ML model. The model will predict multipath delays, which can then be used to correct incoming GNSS measurements, thereby improving positioning accuracy. Our objective is to identify optimal configurations for ML models that deliver precise predicted code multipath delay corrections, thus enhancing GNSS positioning accuracy.
本项目旨在应用机器学习（ML）技术对全球导航卫星系统代码测量中的代码多径延迟进行建模。将从双频 GNSS 测量中计算这些延迟，并选择与多径延迟相关的变量作为训练 ML 模型的输入。该模型将预测多径延迟，然后可用于校正输入的 GNSS 测量值，从而提高定位精度。我们的目标是确定 ML 模型的最佳配置，以提供 精确的预测 代码 多径延迟 校正 、从而提高 GNSS 定位精度。

Answerable research questions for 3-5 students
可供 3-5 名学生回答的研究问题

What are the traditional methods for correcting multipath effects?
校正 multipatheffect 的传统方法是什么？

How to obtain the multipath delay time series from dual-frequency GNSS measurements?
如何从双频全球导航卫星系统测量中获取多径延迟时间序列?

What variables are related to the multipath delays and can be used as inputs for training the machine learning models? How the machine learning models’ performance varies when using different input variables combinations?
哪些变量与多路径延迟有关并可用作训练机器学习模型的输入？ 机器学习 模型的性能在使用不同的输入变量组合时有何不同？

What modelling and prediction precisions can be achieved through the trained machine learning models
通过训练有素的机器学习通过训练有素的机器学习模型.

What positioning performance improvements (measured by accuracy or Precise Point Positioning [PPP] convergence time) can be achieved by introducing multipath corrections generated by machine learning models?
通过引入机器学习模型生成的多径修正，可以实现哪些定位性能改进（以精度或精确点定位 [PPP] 收敛时间衡量）？

3-5 key references (very preferable for students to start)
3-5 篇主要参考文献（学生最好能从这些参考文献入手）

Braasch, M. S. (2017). Multipath. In (pp. 443-468). Springer International Publishing. https://doi.org/10.1007/978-3-319-42928-1_15
Braasch, M. S. (2017).Multipath.第 443-468 页）。Springer International Publishing.https://doi.org/10.1007/978-3-319-42928-1_15

Strode, P. R. R., & Groves, P. D. (2016). GNSS multipath detection using three-frequency signal-to-noise measurements. GPS solutions, 20(3), 399-412. https://doi.org/10.1007/s10291-015-0449-1
Strode, P. R. R., & Groves, P. D. (2016)。使用三频信噪比测量的 GNSS 多径检测。GPS solutions, 20(3), 399-412. https://doi.org/10.1007/s10291-015-0449-1

Li, Q., Xia, L., Chan, T. O., Xia, J., Geng, J., Zhu, H., & Cai, Y. (2020). Intrinsic Identification and Mitigation of Multipath for Enhanced GNSS Positioning. Sensors, 21(1), 188. https://doi.org/10.3390/s21010188
Li, Q., Xia, L., Chan, T. O., Xia, J., Geng, J., Zhu, H., & Cai, Y. (2020).增强型 GNSS 定位的多径本征识别与缓解。Sensors, 21(1), 188. https://doi.org/10.3390/s21010188

Bilich, A., Larson, K. M., & Axelrad, P. (2008). Modeling GPS phase multipath with SNR: Case study from the Salar de Uyuni, Boliva. Journal of Geophysical Research, 113(B4). https://doi.org/10.1029/2007jb005194
Bilich, A., Larson, K. M., & Axelrad, P. (2008).Modeling GPS phase multipath with SNR: Case study from the Salar de Uyuni, Boliva.Journal of Geophysical Research, 113(B4). https://doi.org/10.1029/2007jb005194

Smyrnaios, M., Schn, S., Liso, M., & Jin, S. (2013). Multipath propagation, characterization and modeling in GNSS. Geodetic sciences-observations, modeling and applications, 99-125.
Smyrnaios, M., Schn, S., Liso, M., & Jin, S. (2013).GNSS中的多径传播、特征描述和建模。

Required major of studies, skills, knowledge, and speciality
必修专业、技能、知识和专长

Students majoring data science and computer science can participate in the project.
数据科学和计算机科学专业的学生可以参与该项目。

Programming skills (Python or Matlab)
编程技能（Python 或 Matlab)

Industry-based project: Student IP Agreement. This is the IP model agreed between the parties. Please note that it is QUT policy that where possible students should be allowed to keep their IP. If students are asked to assign their work then please provide a brief rationale as additional permissions are needed by QUT to approve.
行业项目：学生知识产权协议。这是双方商定的知识产权模式。请注意，昆士兰科技大学的政策是尽可能允许学生保留自己的知识产权。如果要求学生转让自己的作品，请提供简要理由 ，因为昆士兰科技大学需要额外的许可才能批准。

☐ Project IP vests in the Student with a license back to Industry Partner (licence)
☐ 项目知识产权归学生所有，并向行业合作伙伴返还许可权(licence)

OR
或

☐ Project IP vests in the Industry Partner with a licence back to the Student (assignment)
☐ 项目知识产权归属于行业合作伙伴，并向学生返还许可权 （转让）

OR
或

☒ Academic project
☒ 学术项目

Number of students
学生人数

3-5

Student names (if known)
学生姓名（如果知道）

1

2

3

4

5

Remarks on conditions of offer
要约条件备注

The supervising team will shortlist the candidates after their application.
申请结束后，督导组将对候选人进行筛选。