The Internet is playing an increasingly crucial role in providing real-time interactive communication for sensors, devices, and systems through applications spanning across consumer, commercial, industrial, infrastructure, and military disciplines. However, many such delay-sensitive applications suffer from poor performance due to bufferbloat (excessive buffering of packets at network routers leading to high queuing delay and large delay variations).

Prof.Steven H.Low     Dr.Ka-Cheong Leung

To address the challenge, a research team led by  Dr. Ka-Cheong Leung, Associate Professor of the School of Computer Science and Technology at the Harbin Institute of Technology, Shenzhen (HITSZ), also a former faculty member at the University of Hong Kong (HKU), in collaboration with Prof. Steven H. Low at the California Institute of Technology (Caltech), developed a novel network router technology – Self-Tuning Controlled Delay (Self-tuning CoDel) – which can stabilize the queueing delays at all bottleneck links, thereby mitigating bufferbloat.

This advanced network router technology has recently published in IEEE/ACM Transactions on Networking.

None of the existing active queue management (AQM) algorithms has been widely deployed because of complicated parameter tuning and sensitivity of algorithm performance to parameter settings. Their performance and stability are unclear in general networks with multiple bottleneck links, such as the Internet. Our research efforts have focused on solving this difficult yet important research problem which has been long standing for more than two decades.

Self-Tuning CoDel can be employed to many current real-time interactive applications with minimal hardware or software modifications at routers and Transmission Control Protocol (TCP) clients. Moreover, it is expected that this technology will help revolutionize the design for future energy-efficient all-optical routers. Taking benefits of this, the team is planning to further upgrade the current Self-Tuning CoDel by making it more scalable and push for standardization at the Internet Engineering Task Force (IETF), as well as partnering with network equipment manufacturers for product realization.

Publication details:

Jiancheng Ye, Ka-Cheong Leung, and Steven H. Low

Combating Bufferbloat in Multi-Bottleneck Networks: Theory and Algorithms

IEEE/ACM Transactions on Networking, Vol. 29, No. 4, pp. 1477-1493, August 2021

DOI: 10.1109/TNET.2021.3066505

Funding details:

Ka-Cheong Leung (PI), Victor On-Kwok Li (Co-I), and Steven H. Low (Co-I)

A Robust Link Pricing Theory for Combating Bufferbloat

General Research Fund, Research Grants Council, Project Code 17204614