About Me

I am a Ph.D. student at Shenzhen International Graduate School, Tsinghua University, supervised by Prof. Shao-Lun Huang. My research focuses on information theory, integrated sensing and communication (ISAC), and machine learning. I am particularly interested in exploring the unified relationships among Shannon information, Fisher information, and conditional variance, with applications to wireless communication systems.

I have been a visiting scholar at the Massachusetts Institute of Technology and University of California, Berkeley, collaborating with leading researchers in the field.

I am currently seeking job opportunities. Feel free to contact me at: yuanxj23@mails.tsinghua.edu.cn

Education

• Shenzhen International Graduate School, Tsinghua University Ph.D. (Sept. 2023 - Jan. 2027, expected)
  Data Science and Information Technology. Advisor: Prof. Shao-Lun Huang
  Leaders of Future Scholarship (Distinguished, Highest Tier, 2025); Tsinghua University Second-class Scholarship (2025)
• Massachusetts Institute of Technology (MIT) Visiting Scholar (Dec. 2024 - Apr. 2025)
  Co-supervised by Prof. Lizhong Zheng
• University of California, Berkeley Visiting Scholar (Sept. - Dec. 2024)
  Co-supervised by Prof. Khalid M. Mosalam
• Shenzhen International Graduate School, Tsinghua University M.S. (Sept. 2020 - Sept. 2023)
  Data Science and Information Technology. Advisors: Prof. Zhi Wang & Prof. Wenwu Zhu
  Excellent Teaching Assistant (2023); Tsinghua University First-class Scholarship (2022)
• Xi'an Jiaotong University B.E. in Information Engineering (Sept. 2016 - June 2020)
  Outstanding Graduates (2020); National Scholarship (2017, 2018); Outstanding Students (2017, 2018); First Prize in Shaanxi Province in Mathematical Contest in Modeling (2018)

Research Interests

• Information Theory and Statistical Inference: Unified relationships among Shannon measures, Fisher information, MMSE, and conditional variance, including de Bruijn and I-MMSE type identities.
• Integrated Sensing and Communication (ISAC): Fundamental tradeoff boundaries between communication rate and sensing utility in bistatic ISAC systems, with optimized input design.
• Network Measurement and System Optimization: End-to-end measurement and analysis of real-world 5G services, with system-level optimization for lower latency and reduced rebuffering.
• Machine Learning and Graph Neural Networks: GNN, transfer learning, and differentiable surrogate modeling for channel estimation and structural safety applications.

Publications and Patents

Journal Articles

1. X. Yuan and K. M. Mosalam, "Prediction of the most fire-sensitive point in building structures with differentiable agents for thermal simulators," Computer-Aided Civil and Infrastructure Engineering, vol. 40, pp. 2584–2611, 2025. [JCR Q1 TOP Journal, Cover Article]
2. X. Yuan, M. Wu, Z. Wang, Y. Zhu, M. Ma, J. Guo, Z.-L. Zhang, and W. Zhu, "Understanding 5G performance for real-world services: A content provider's perspective," IEEE/ACM Transactions on Networking, vol. 33, no. 4, pp. 1746–1761, Aug. 2025. [CCF-A]

Conference Papers

1. X. Yuan, T. Peng, Z. Liu, and S.-L. Huang, "On the equivalence relationships among Fisher information, Shannon measures and variance," in Proc. 2026 IEEE Int. Symp. Inf. Theory (ISIT), 2026, under review.
2. X. Yuan*, T. Peng*, H. Fu, and S.-L. Huang, "Multi-letter analysis of marginal rate-Fisher information tradeoff: Unbounded gain," in Proc. 2025 Asia Pac. Workshop Data Sci. Inf. Theory (APWDSIT), 2025, to appear.
3. X. Yuan, T. Peng, H. Fu, and S.-L. Huang, "Unbounded multi-letter marginal gain for bistatic integrated sensing and communication," in Proc. 2025 Asia Pac. Workshop Data Sci. Inf. Theory (APWDSIT), 2025, to appear.
4. X. Yuan*, J. Wang*, R. Chen, Z. Liu, J. Song, S.-L. Huang, and X. Guan, "Enhancing VLC vehicle networks in channel estimation, coding, and multiple access," in Proc. 2024 IEEE Globecom Workshops, 2024.
5. X. Yuan, M. Wu, Z. Wang, Y. Zhu, M. Ma, J. Guo, Z.-L. Zhang, and W. Zhu, "Understanding 5G performance for real-world services: A content provider's perspective," in Proc. ACM SIGCOMM 2022 Conf., New York, NY, USA, Aug. 2022, pp. 101–113. [CCF-A, First from Tsinghua SIGS]

Patents

1. X. Yuan and Q. Du, "A resource allocation method for edge computing based on priority and collaboration," Patent ZL 2020 1 0473969.6, authorized June 28, 2024.
2. X. Yuan and Q. Du, "A resource allocation method for edge computing," Patent ZL 2020 1 0460707.6, authorized March 29, 2024.

Research Experiences

• Equivalence relationships among Fisher information, Shannon measures and variance (2025–2026)

  Proposed the Equivalent Polynomial Representation framework in double-scaling Gaussian MIMO channels, unifying Shannon mutual information, differential entropy, Fisher information, MMSE, and conditional variance. Derived extended de Bruijn's identity and I-MMSE relations. Submitted to ISIT 2026.

• Theoretical tradeoff boundaries in integrated sensing and communication (2025)

  Studied marginal tradeoffs between communication and sensing in bistatic ISAC systems. Achieved exponential growth in sensing performance (error exponent metric) with negligible communication rate loss through optimized input distribution perturbations. Published in APWDSIT 2025.

• Differentiable graph neural networks and transfer learning applications (2024–2025)

  Interdisciplinary research on predicting fire-sensitive points in building structures. Integrated GNN with transfer learning using differentiable surrogate models. Achieved over three orders of magnitude speedup compared to traditional simulations. Published as cover article in CACAIE (JCR Q1 TOP).

• Channel estimation using neural network with detection feedback (2023–2024)

  Addressed underutilization of data bit information in pilot-based channel estimation by proposing detection feedback with neural networks for secondary estimation. Published in Globecom 2024 Workshops.

• Measurement study on 5G networks for real-world services (2021–2023)

  Conducted comprehensive 5G network measurements from content provider's perspective, measuring end-to-end, RAN, and core network performance across SA/NSA 5G and other networks. Proposed dynamic caching strategy, reducing rebuffer rate by up to 7%. Published in SIGCOMM 2022 and IEEE/ACM Transactions on Networking.

• Edge computing resource allocation based on reinforcement learning (2019–2022)

  Used DQN with fixed Q-target and experience replay for computational resource allocation. Applied DDPG for cloud-edge collaboration scenarios. Two patents granted.

Teaching and Industry Experience

• Teaching Assistant for Distributed Machine Learning (Spring 2023)

  Wrote the experimental manual and conducted laboratory sessions as primary instructor. Received the Excellent Teaching Assistant Award from Tsinghua University.

• Research Intern at Kuaishou Technology (2021–2023)

  Conducted 5G network measurements and optimization from content provider's perspective. Proposed dynamic caching strategy, reducing SA 5G live streaming rebuffer rate by up to 7% in 18-day experiment with ~9 million users. Published in SIGCOMM 2022 (first from Tsinghua SIGS) and IEEE/ACM Transactions on Networking.

• Visiting Student at Tokyo University of Science, Takemura Lab (Aug. - Sept. 2018)

  Japan-Asia Youth Exchange Program in Science (Sakura Science Program). Developed VR game using Unity and C# with panoramic camera, treadmill, and VR equipment.

• Summer Program in Big Data Community Detection, National University of Singapore (July - Aug. 2018)

  Completed city community detection project using travel blogs and images with TF/IDF algorithm and places365 model. Team ranked 1st among 11 groups with A+ grade.

• Director of Back-end Group in eeyes.net (2017–2019)

  A student technical community at Xi'an Jiaotong University. Built websites with PHP and Laravel framework.

Contact

Email: yuanxj23@mails.tsinghua.edu.cn / cantjie@163.com