Mengying Zhou (周孟莹)
Mengying Zhou (周孟莹)

Fourth-year Ph.D. Candidate at Fudan University

github
linkedin

About Me

I am a fourth-year Ph.D. candidate in the Mobile Systems and Networking (MSN) Group (opens new window) of the SONIC Laboratory, School of Computer Science at Fudan University, advised by Prof. Xin Wang (opens new window) and Prof. Yang Chen (opens new window). My research interests include Machine Learning for Networking Systems (NetAI), Next-Generation Internet Architecture, and Social Network Analysis.

I graduated from Lanzhou University in 2019 with a B.S. in Computer Science.

News

  • 📌 [09/2022] One paper titled "Artemis: A Latency-Oriented Naming and Routing System" got accepted in IEEE Transactions on Parallel and Distributed Systems (TPDS).
  • 📌 [08/2022] Our "Locating CDN Edge Servers with HTTP Responses" won the first place in the SIGCOMM 2022 Student Research Competition (opens new window)!
  • 📌 [06/2022] One paper titled "Locating CDN Edge Servers with HTTP Responses" got accepted in Proc. of the SIGCOMM '22 Posters and Demos.
  • 📌 [03/2022] One paper titled "FlexHTTP: An Intelligent and Scalable HTTP Version Selection System" got accepted in Proc. of the 2nd workshop on Machine Learning and Systems (EuroMLSys), co-located with EuroSys'22.
  • 📌 [11/2021] One paper titled "Polygon: A QUIC-Based CDN Server Selection System Supporting Multiple Resource Demands" got accepted in Proc. of the 22nd ACM/IFIP Middleware Conference, Industry Track.

Education

Research Interests

  • Machine Learning for Networking Systems (NetAI)
  • Next-Generation Internet Architecture
  • Social Network Analysis

Publications

  • FlexHTTP: An Intelligent and Scalable HTTP Version Selection System. Proc. of the 2nd workshop on Machine Learning and Systems (EuroMLSys), co-located with EuroSys'22.

    Mengying Zhou, Zheng Li, Shihan Lin, Xin Wang, and Yang Chen.

Our work find that under different network conditions and web page structures, neither HTTP/2 nor HTTP/3 can always perform better. Motivated by this finding, we propose FlexHTTP, an intelligent and scalable HTTP version selection system. The proof-of-concept evaluation demonstrates that FlexHTTP almost achieves an optimal HTTP version selection with a hit ratio of 0.97. When compared with the ideal results, FlexHTTP only displays a small gap with a value of 30ms median Speed Index.

[PDF] [Slide]

  • Polygon: A QUIC-Based CDN Server Selection System Supporting Multiple Resource Demands. Proc. of the 22nd ACM/IFIP Middleware Conference (Middleware’21), Industry Track, Virtual Event, Canada, Dec. 2021.

    Mengying Zhou, Tiancheng Guo, Yang Chen, Junjie Wan, and Xin Wang.

Our work find that requests in different application scenarios would have different resource type priorities. And motivated by this finding, we propose Polygon, a QUIC-based CDN server selection system supporting multiple resource demands. It shows a significant improvement in job completion time and resource utilization efficiency.

[PDF] [Slides]

Determining the physical locations of CDN Points of Presence (PoPs) is fundamental to understanding and diagnosing CDN services. Yet, the popular deployment of IP Anycast in CDNs has rendered existing geolocation tools unreliable. To fill this gap, we present an HTTP-based solution that leverages subtle geographic hints in HTTP responses to locate CDN PoPs at the city-level granularity. The evaluation shows that our technique achieves over 90% accuracy with an average error distance of less than 40 km.

[PDF] [Video (opens new window)] [Demo (opens new window)] [Code (opens new window)]

  • Artemis: A Latency-Oriented Naming and Routing System. Xuebing Li, Yang Chen, Mengying Zhou, Tiancheng Guo, Chenhao Wang, Yu Xiao, Junjie Wan, and Xin Wang. To appear: IEEE Transactions on Parallel and Distributed Systems (TPDS).

  • Deep Learning for POI Recommendation: A Survey. Jiaxin Tang, Yang Chen, Mengying Zhou, and Xin Wang. Computer Engineering, 2021, 1000-3428.0061598. [PDF]

  • Internet Data Transfer Protocol QUIC: A Survey. Xuebing Li, Yang Chen, Mengying Zhou, and Xin Wang. Journal of Computer Research and Development, 2020, 57(9):1864-1876. [PDF] [Slides]

Research Experience

  • Machine Learning-based Network System with Effectiveness and EfficiencyPresent

    Advisor: Prof. Yang Chen (opens new window), School of Computer Science, Fudan University.

    • Analysis find that different network conditions and web page structures affect the transmission performance difference between QUIC and TCP.
    • Responsible for building a machine learning-based model to select the optimal protocol under different network conditions and web content.
    • Experimental results show that the average page load time of the adaptive hybrid protocol system is shorter than that of the single protocol system.

  • User Awareness-based Next-generation Network StructurePresent

    Advisor: Prof. Yang Chen (opens new window), School of Computer Science, Fudan University.

    • It is necessary to dynamically adjust the network conditions according to user needs.
    • Responsible for building an selection system that provides the most suitable network parameters.
    • Emulation results show that meeting dynamic needs can improve system resource utilization.

  • Understanding the Behavioral Differences Between Users with Different Loyalty on AirbnbPresent

    Advisors: Prof. Yang Chen (opens new window), School of Computer Science, Fudan University. Prof. Bruno Abrahao (opens new window), Center for Business Education and Research, New York University Shanghai.

    • Analysis find that there are behavior differences between users with different loyalty.
    • Responsible for establishing a deep learning model, DeepChurn, that predicts whether a user is a loyal user based on the user's behavior preference.
    • DeepChurn's prediction performance is superior than the SOTA models' 0.2 accuracy.

  • Data-Driven Management Platform for Social Non-profit Organizations2019.6 - 2022.02

    Advisor: Prof. Jar-der Luo (opens new window), Department of Sociology, Tsinghua University.

    • Provide electronic management tools for non-profit social organizations.
    • Responsible for the development of WeChat mini programs and an online system for analyzing the activity patterns of members of social organizations.
    • This system has been deployed and used on a large scale in Chengdu.

→ [Full list]

Awards & Honors

Extracurricular Experience