Junting CHEN

Adventure in Scottish Highlands 

Ph.D (HKUST), B.Sc (NJU)
Assistant Professor

The Chinese University of Hong Kong, Shenzhen
School of Science and Engeering, RA 215
Shenzhen, Guangdong 518172, China

E-mail: juntingc@cuhk.edu.cn
Tel: +86 (755) 235 19652

“I know of no better life purpose than to perish in attempting the great and the impossible…”  —  Nietzsche, 1873.

Opening for PhD students / postdocs / visiting positions: I am actively looking for PhD students with full scholarship to be admitted in the Computer and Information Engineering (CIE) program at CUHK-SZ. The program duration is typically 4-5 years. I also have several funded positions for postdocs, visisting students, and research assistants. See details here …


Junting is an Assistant Professor with the School of Science and Engeering in The Chinese University of Hong Kong, Shenzhen (CUHK-SZ). Prior to joining CUHK-SZ in Feburary 2019, he was a postdoctoral research associate in Prof. Urbashi Mitra's group in University of Southern California, Los Angeles, USA, from November 2016 to January 2019, and a postdoctoral research fellow in Prof. David Gesbert's group in EURECOM, France, from 2015 to 2016.

Junting received the B.Sc degree in Electronic Science and Engineering from Nanjing University (NJU), Nanjing, China, in 2009, and the Ph.D. degree in Electronic and Computer Engineering from the Hong Kong University of Science and Technology (HKUST) , Hong Kong, in 2015. He was advised by Prof. Vincent Lau. From February to December 2014, he visited the Laboratory for Information & Decision Systems (LIDS), Massachusetts Institute of Technology (MIT) , USA, and was advised by Prof. Moe Z. Win.

Research Highlights

His research encompasses a broad range of problems in signal processing, optimizations, stochastic approximation and machine learning with applications to wireless communications and localization. His most recent work include sparse signal processing for underwater acoustic communication and localization, machine learning and optimization for unmanned aerial vehicles (UAVs) aided communication networks, device-to-device (D2D) assisted cellular communications, massive MIMO systems, and cooperative localization.

3D Communication Network Formation

  • Can we optimally place a flying relay base station (BS) to help the communication between ground terminals in arbitrary obstructive environments?

  • Can we find the optimal relay position in limited search trajectory, if we have no prior information of the signal-blocking obstacles?

To answer these questions, we need to start from modeling, and then incorporate the techniques from learning to optimization.
energy samples for source localization 

Somewhat surprisingly, we find affirmative answers from our preliminary result! We are able to find the optimal relay position in 3D to relay the signal from a BS to an obstructive ground user, no matter how complex the obstructive environment is! Moreover, the search trajectory length is only linear to the BS-user distance, even though we do not have any prior information of the environment.

I am interested in optimal 3D communication network formation to both avoiding propagation blockage to intended ground users and, at the same time, exploiting propagation blockage to isolate the signal to unintended users. I believe that such a new frontier will also be a promising enabler to other fast developing technologies, such as millimeter-wave (mmWave) which highly prefers line-of-sight (LOS) propagation conditions. [See more …]

Machine Learning for Network Localization and Communication

  • How to localize a signal source without knowing the propagation model?

  • How to localize an attacker in a sensor network by analyzing the packet delivery ratio?

I believe that the solution concepts to the above two questions are related, and there are common structural properties to be learned and exploited. In general, I am interested in the learning and estimation (e.g., a source location, propagation environment) with the help of an ample of (yet, low resolution) sensing nodes, while the underlying source model is unknown (or too complicated to acquire). This will lead to important applications in future IoT (Internet-of-things) networks powered by low-cost long-duration nodes.

In particular, in underwater source localization, we have developed matrix/tensor observation models and used matrix/tensor decomposition techniques to extract the source signature vectors for localization. While we make no assumption on the energy propagation model, the method is also robust to various imperfection such as sensor location uncertainty. [See more …]


2001 Longxiang Ave, CUHK-SZ, RA 215
Longgang District, Shenzhen
Guangdong 518172, China

E-mail: juntingc@cuhk.edu.cn
Tel: +86 (755) 235 19652