Qichen Su
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Qichen Su has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Jiawei Huang
Central South University
7 shared publications - Jianxin Wang
School of Computer Science and Engineering, Central South University, Changsha, Hunan, China
6 shared publications - Weihe Li
University of Edinburgh, Edinburgh, United Kingdom
6 shared publications - Zhaoyi Li
Central South University, Changsha, China
4 shared publications - Jin Ye
School of Computer and Electronic Information, Guangxi University, Nanning, China
2 shared publications - Jingling Liu
Central South University, Changsha, China
2 shared publications
Latest Publications
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2754-2765
L asso : Accurate and Efficient Detection of Long-Lived Sparse Items in High-Speed Data Streams
In high-speed data streams, identifying long-lived (also referred to as persistent) sparse items is critical, as such patterns may indicate stealthy or low-rate threats yet remain largely underexplored. Although recent studies have begun to examine this problem, existing approaches either suffer from low lookup accuracy due to coarse update strategies or rely on complex data structures with costly update operations, overlooking the practical requirement of deployability. These limitations hinder scalability, particularly as programmable switches and FPGAs are increasingly adopted as data-processing substrates that sustain high-speed processing under strict resource and operational constraints. To address these challenges, we propose Lasso, a lightweight and hardware-conscious approach that achieves high detection accuracy under tight memory budgets while sustaining high processing throughput on industry-grade hardware, including Tofino-1 programmable switches and FPGA platforms. Lasso leverages the observation that long-lived sparse items exhibit a small gap between persistence and frequency, evicting items with large deviations to prioritize promising candidates. In addition, Lasso incorporates fine-grained, temporally aware protection to prevent long-lived items from being prematurely displaced by abundant short-lived items in highly skewed data streams. We further develop a formal analytical model to establish the theoretical soundness of Lasso. Extensive evaluations across CPU, Tofino, and FPGA platforms demonstrate that Lasso delivers high accuracy and throughput while operating within strict resource constraints.
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2754-2765
Open AccessL asso : Accurate and Efficient Detection of Long-Lived Sparse Items in High-Speed Data Streams
In high-speed data streams, identifying long-lived (also referred to as persistent) sparse items is critical, as such patterns may indicate stealthy or low-rate threats yet remain largely underexplored. Although recent studies have begun to examine this problem, existing approaches either suffer from low lookup accuracy due to coarse update strategies or rely on complex data structures with costly update operations, overlooking the practical requirement of deployability. These limitations hinder scalability, particularly as programmable switches and FPGAs are increasingly adopted as data-processing substrates that sustain high-speed processing under strict resource and operational constraints. To address these challenges, we propose Lasso, a lightweight and hardware-conscious approach that achieves high detection accuracy under tight memory budgets while sustaining high processing throughput on industry-grade hardware, including Tofino-1 programmable switches and FPGA platforms. Lasso leverages the observation that long-lived sparse items exhibit a small gap between persistence and frequency, evicting items with large deviations to prioritize promising candidates. In addition, Lasso incorporates fine-grained, temporally aware protection to prevent long-lived items from being prematurely displaced by abundant short-lived items in highly skewed data streams. We further develop a formal analytical model to establish the theoretical soundness of Lasso. Extensive evaluations across CPU, Tofino, and FPGA platforms demonstrate that Lasso delivers high accuracy and throughput while operating within strict resource constraints.
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6944-6955
Cardinality is Not Enough: Super Host Detection via Segmented Cardinality Estimation
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 6944-6955
Open AccessCardinality is Not Enough: Super Host Detection via Segmented Cardinality Estimation
- 3 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 2103-2118
Toward QoE-Fairness for Video Streaming Over Heterogeneous Networks: An Innovative Bandwidth Allocation Mechanism
With the growing ubiquity of video streaming, ensuring a fair and high quality of experience (QoE) for users has emerged as a shared concern among video content providers. State-of-the-art video delivery systems achieve QoE fairness through bottleneck bandwidth allocation across multiple video streams, all based on the assumption of a unified congestion control (CC) protocol. However, the widespread use of heterogeneous CC protocols on the Internet not only disrupts QoE fairness among video streams but also poses challenges in achieving fast convergence under dynamic bandwidth. To address these issues, we propose a QoE-Fairness aware bandwidth allocation mechanism called Fabam, which establishes a unified QoE control plane across heterogeneous CC protocols. Fabam constructs independent virtual targets based on the real-time QoE of each video stream to achieve QoE fairness, and offers rapid convergence for the underlying CC protocols to improve efficiency. In addition, we propose a Deep Neural Network (DNN)-based multi-step mapping model aimed at balancing the performance and overhead of Fabam, thereby enhancing its deployment potential in practical applications. We implement Fabam on QUIC and integrate it with Dash.js. The evaluation results demonstrate the significant superiority of Fabam over the state-of-the-art approaches, including an enhancement of 44.01% in QoE fairness and an improvement of 36.39% in QoE efficiency. Meanwhile, Fabam-DNN maintains satisfactory QoE fairness while supporting multiple users at a low cost.
- 3 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 2103-2118
Toward QoE-Fairness for Video Streaming Over Heterogeneous Networks: An Innovative Bandwidth Allocation Mechanism
With the growing ubiquity of video streaming, ensuring a fair and high quality of experience (QoE) for users has emerged as a shared concern among video content providers. State-of-the-art video delivery systems achieve QoE fairness through bottleneck bandwidth allocation across multiple video streams, all based on the assumption of a unified congestion control (CC) protocol. However, the widespread use of heterogeneous CC protocols on the Internet not only disrupts QoE fairness among video streams but also poses challenges in achieving fast convergence under dynamic bandwidth. To address these issues, we propose a QoE-Fairness aware bandwidth allocation mechanism called Fabam, which establishes a unified QoE control plane across heterogeneous CC protocols. Fabam constructs independent virtual targets based on the real-time QoE of each video stream to achieve QoE fairness, and offers rapid convergence for the underlying CC protocols to improve efficiency. In addition, we propose a Deep Neural Network (DNN)-based multi-step mapping model aimed at balancing the performance and overhead of Fabam, thereby enhancing its deployment potential in practical applications. We implement Fabam on QUIC and integrate it with Dash.js. The evaluation results demonstrate the significant superiority of Fabam over the state-of-the-art approaches, including an enhancement of 44.01% in QoE fairness and an improvement of 36.39% in QoE efficiency. Meanwhile, Fabam-DNN maintains satisfactory QoE fairness while supporting multiple users at a low cost.
- 26 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 1838-1850
SIM: Accelerating Distributed DNN Training by Exploring Gradient Similarity
Synchronous stochastic gradient descent (SSGD) has been widely used in distributed deep learning. However, since the local gradients need to be shared among workers at every iteration, SSGD performance is significantly influenced by network bottlenecks caused by either heterogeneous environment or bandwidth contention. To solve this problem, asynchronous parallel (ASP) strategy allows each worker to update parameters independently without synchronization, while suffering from accuracy loss and convergence inefficiency. In this paper, we propose a novel similarity-based synchronization scheme called SIM, which mitigates the impact of network bottlenecks and ensures convergence efficiency. Specifically, SIM reduces the number of aggregation workers based on the gradient similarity between global and local gradients, therefore shrinking the waiting time for the stragglers. We provide a theoretical analysis of convergence efficiency and conduct large-scale testbed experiments on CIFAR-10 and SQUAD dataset. The experimental results show that SIM reduces the convergence time of four classical deep learning models by up to 40%.
- 26 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 1838-1850
SIM: Accelerating Distributed DNN Training by Exploring Gradient Similarity
Synchronous stochastic gradient descent (SSGD) has been widely used in distributed deep learning. However, since the local gradients need to be shared among workers at every iteration, SSGD performance is significantly influenced by network bottlenecks caused by either heterogeneous environment or bandwidth contention. To solve this problem, asynchronous parallel (ASP) strategy allows each worker to update parameters independently without synchronization, while suffering from accuracy loss and convergence inefficiency. In this paper, we propose a novel similarity-based synchronization scheme called SIM, which mitigates the impact of network bottlenecks and ensures convergence efficiency. Specifically, SIM reduces the number of aggregation workers based on the gradient similarity between global and local gradients, therefore shrinking the waiting time for the stragglers. We provide a theoretical analysis of convergence efficiency and conduct large-scale testbed experiments on CIFAR-10 and SQUAD dataset. The experimental results show that SIM reduces the convergence time of four classical deep learning models by up to 40%.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (5) , 6979-6990
Variable-Sized Tiling and Rate Allocation for 360° Video Streaming
In recent years, the quality of 360 ${}^{\circ }$ video over bandwidth-constrained networks has received a lot of attention. To reduce the bandwidth cost, some tile-based 360 ${}^{\circ }$ video streaming systems split video spatially into variable-sized coarse-grained tiles by grouping areas with the similar quality levels. However, these approaches do not minimize the number of tiles, resulting in considerable overhead in re-encoding the boundaries between small tiles. To improve the coding efficiency of 360 ${}^{\circ }$ video, we propose a new video tiling scheme called TRAS, by using the convex decomposition of concave polygon. Meanwhile, to maximize user's quality of experience (QoE), we propose a greedy bitrate selection algorithm by considering the tile size and coding efficiency. The experiment results demonstrate that, compared with the state-of-the-art approaches, TRAS reduces the coding overhead by about 18% and significantly improves the video quality by about 5 dB.
- 1 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (5) , 6979-6990
Variable-Sized Tiling and Rate Allocation for 360° Video Streaming
In recent years, the quality of 360 ${}^{\circ }$ video over bandwidth-constrained networks has received a lot of attention. To reduce the bandwidth cost, some tile-based 360 ${}^{\circ }$ video streaming systems split video spatially into variable-sized coarse-grained tiles by grouping areas with the similar quality levels. However, these approaches do not minimize the number of tiles, resulting in considerable overhead in re-encoding the boundaries between small tiles. To improve the coding efficiency of 360 ${}^{\circ }$ video, we propose a new video tiling scheme called TRAS, by using the convex decomposition of concave polygon. Meanwhile, to maximize user's quality of experience (QoE), we propose a greedy bitrate selection algorithm by considering the tile size and coding efficiency. The experiment results demonstrate that, compared with the state-of-the-art approaches, TRAS reduces the coding overhead by about 18% and significantly improves the video quality by about 5 dB.
- 28 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 23 (12) , 12280-12297
Optimizing Video Streaming in Dynamic Networks: An Intelligent Adaptive Bitrate Solution Considering Scene Intricacy and Data Budget
Adaptive Bitrate (ABR) algorithms have become increasingly important for delivering high-quality video content over fluctuating networks. Considering the complexity of video scenes, video chunks can be separated into two categories: those with intricate scenes and those with simple scenes. In practice, it has been observed that improving the quality of intricate chunks yields more substantial improvements in Quality of Experience (QoE) compared with focusing solely on simple chunks. However, the current ABR schemes either treat all chunks equally or rely on fixed linear-based reward functions, which limits their ability to meet real-world requirements. To tackle these limitations, this paper introduces a novel ABR approach called CAST (Complex-scene Aware bitrate algorithm via Self-play reinforcemenT learning), which considers the scene complexity and formulates the bitrate adaptation task as an explicit objective. Leveraging the power of parallel computing with multiple agents, CAST trains a neural network to achieve superior video playback quality for intricate scenes while minimizing playback freezing time. Moreover, we also introduce a new variant of our proposed approach called CAST-DU, to address the critical issue of efficiently managing users’ limited cellular data budgets while ensuring a satisfactory viewing experience. Furthermore, we present CAST-Live, tailored for live streaming scenarios with constrained playback buffers and considerations for energy costs. Extensive trace-driven evaluations and subjective tests demonstrate that CAST, CAST-DU, and CAST-Live outperform existing off-the-shelf schemes, delivering a superior video streaming experience over fluctuating networks while efficiently utilizing data resources. Moreover, CAST-Live demonstrates effectiveness even under limited buffer size constraints while incurring minimal energy costs.
- 28 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 23 (12) , 12280-12297
Optimizing Video Streaming in Dynamic Networks: An Intelligent Adaptive Bitrate Solution Considering Scene Intricacy and Data Budget
Adaptive Bitrate (ABR) algorithms have become increasingly important for delivering high-quality video content over fluctuating networks. Considering the complexity of video scenes, video chunks can be separated into two categories: those with intricate scenes and those with simple scenes. In practice, it has been observed that improving the quality of intricate chunks yields more substantial improvements in Quality of Experience (QoE) compared with focusing solely on simple chunks. However, the current ABR schemes either treat all chunks equally or rely on fixed linear-based reward functions, which limits their ability to meet real-world requirements. To tackle these limitations, this paper introduces a novel ABR approach called CAST (Complex-scene Aware bitrate algorithm via Self-play reinforcemenT learning), which considers the scene complexity and formulates the bitrate adaptation task as an explicit objective. Leveraging the power of parallel computing with multiple agents, CAST trains a neural network to achieve superior video playback quality for intricate scenes while minimizing playback freezing time. Moreover, we also introduce a new variant of our proposed approach called CAST-DU, to address the critical issue of efficiently managing users’ limited cellular data budgets while ensuring a satisfactory viewing experience. Furthermore, we present CAST-Live, tailored for live streaming scenarios with constrained playback buffers and considerations for energy costs. Extensive trace-driven evaluations and subjective tests demonstrate that CAST, CAST-DU, and CAST-Live outperform existing off-the-shelf schemes, delivering a superior video streaming experience over fluctuating networks while efficiently utilizing data resources. Moreover, CAST-Live demonstrates effectiveness even under limited buffer size constraints while incurring minimal energy costs.