Xiangyu Gao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiangyu Gao has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Ke Xu
Tsinghua University, Beijing, China
4 shared publications - Tong Li
Renmin University of China, Beijing, China
4 shared publications - Zhuotao Liu
Tsinghua University, Beijing, China
3 shared publications - Hanlin Huang
Department of Computer Science and Technology, Tsinghua University, Beijing, China
2 shared publications - Su Yao
Beijing National Research Center for Information Science and Technology (BNRist), Tsinghua University, Beijing, China
2 shared publications Xinle Du
Huawei Technologies Company Ltd., Shenzhen, China
2 shared publications
Latest Publications
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 6859-6874
An FPGA Integrated Programmable Switch Architecture for Data Plane DNN Inference
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 6859-6874
An FPGA Integrated Programmable Switch Architecture for Data Plane DNN Inference
- Open Access
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 692-706
Pegasus: A Universal Framework for Scalable Deep Learning Inference on the Dataplane
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 692-706
Open AccessPegasus: A Universal Framework for Scalable Deep Learning Inference on the Dataplane
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 33 (5) , 2507-2522
Revisiting Random Early Detection Tuning for High-Performance Datacenter Networks
Random Early Detection (RED) has been integrated into datacenter switches as a fundamental Active Queue Management (AQM) for decades. The accurate configuration of RED parameters is crucial to achieving high throughput and low latency. However, due to the highly dynamic nature of workloads in datacenter networks, maintaining consistently high performance with statically configured RED thresholds poses a challenge. Prior work applies reinforcement learning to predict proper thresholds, but their real-world deployment has been hindered by poor tail performance caused by instability. In this paper, we propose $\textsf {PRED}$ , a novel system that enables automatic and stable RED parameter adjustment in response to traffic dynamics. Specifically, the system employs a Multiplicative-Increase Multiplicative-Decrease (MIMD) strategy to dynamically adapt to flow concurrency while utilizing an Additive-Increase Additive-Decrease (AIAD) mechanism to adapt to flow distribution. We perform extensive evaluations on our physical testbed and large-scale simulations. The results demonstrate that $\textsf {PRED}$ can keep up with the real-time network dynamics generated by realistic workloads. For instance, compared with the static-threshold-based methods, $\textsf {PRED}$ keeps 66% shorter switch queue length and obtains up to 80% lower Flow Completion Time (FCT). Compared with the state-of-the-art learning-based method, $\textsf {PRED}$ reduces the tail FCT by 34%.
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 33 (5) , 2507-2522
Revisiting Random Early Detection Tuning for High-Performance Datacenter Networks
Random Early Detection (RED) has been integrated into datacenter switches as a fundamental Active Queue Management (AQM) for decades. The accurate configuration of RED parameters is crucial to achieving high throughput and low latency. However, due to the highly dynamic nature of workloads in datacenter networks, maintaining consistently high performance with statically configured RED thresholds poses a challenge. Prior work applies reinforcement learning to predict proper thresholds, but their real-world deployment has been hindered by poor tail performance caused by instability. In this paper, we propose $\textsf {PRED}$ , a novel system that enables automatic and stable RED parameter adjustment in response to traffic dynamics. Specifically, the system employs a Multiplicative-Increase Multiplicative-Decrease (MIMD) strategy to dynamically adapt to flow concurrency while utilizing an Additive-Increase Additive-Decrease (AIAD) mechanism to adapt to flow distribution. We perform extensive evaluations on our physical testbed and large-scale simulations. The results demonstrate that $\textsf {PRED}$ can keep up with the real-time network dynamics generated by realistic workloads. For instance, compared with the static-threshold-based methods, $\textsf {PRED}$ keeps 66% shorter switch queue length and obtains up to 80% lower Flow Completion Time (FCT). Compared with the state-of-the-art learning-based method, $\textsf {PRED}$ reduces the tail FCT by 34%.
- 17 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Networking
- Vol. 33 (1) , 210-225
DiffECN: Differential ECN Marking for Datacenter Networks
ECN marking has been integrated into datacenter switches to enable high-throughput and low-latency transport. We observe that current marking schemes are coarse-grained: they blindly mark all flows when congestion occurs, causing large flows to occupy undeserved bandwidth and preventing newly arriving small flows from finishing quickly. In this paper, we propose, a differential marking strategy that marks only the flows that are the culprits of congestion and protects the remaining flows from being limited. We have implemented it in the Barefoot Tofino switch and performed extensive evaluations via both physical testbed and large-scale simulations. The results show that can restrain flows responsible for congestion successfully while providing desirable network performance. For instance, compared to the legacy way of ECN marking, achieves up to 32.5% (40.1%) lower average (99th percentile) flow completion time (FCT) for small flows while delivering similar FCT for large flows under production workloads.
- 17 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Networking
- Vol. 33 (1) , 210-225
DiffECN: Differential ECN Marking for Datacenter Networks
ECN marking has been integrated into datacenter switches to enable high-throughput and low-latency transport. We observe that current marking schemes are coarse-grained: they blindly mark all flows when congestion occurs, causing large flows to occupy undeserved bandwidth and preventing newly arriving small flows from finishing quickly. In this paper, we propose, a differential marking strategy that marks only the flows that are the culprits of congestion and protects the remaining flows from being limited. We have implemented it in the Barefoot Tofino switch and performed extensive evaluations via both physical testbed and large-scale simulations. The results show that can restrain flows responsible for congestion successfully while providing desirable network performance. For instance, compared to the legacy way of ECN marking, achieves up to 32.5% (40.1%) lower average (99th percentile) flow completion time (FCT) for small flows while delivering similar FCT for large flows under production workloads.