HOMEPAGE
ABOUT
I am now a Lecturer in Department of Computer Science and Technology in SooChow University, and working in the team leaded by Prof. He Huang. Before that, I obtained my Ph.D. degree in Department of Computer Science and Technology in Nanjing University, Nanjing in 2024, advised by Prof. Sanglu Lu and Sheng Zhang.
RESEARCH INTERESTS
Focusing on optimizing edge video inference systems, my research aims to achieve low-latency, low-power consumption, and high-performance video inference in resource-constrained edge environments. Specifically, my research encompasses the following three core sub-directions:
Adaptive Video Configuration: In edge computing environments, due to hardware limitations (such as computational power, memory, and battery life), network bandwidth uncertainty, and video content diversity, traditional fixed-configuration methods pose significant challenges to the real-time performance and accuracy of video inference. Thus, I explore dynamically adjusting video configurations (such as resolution, frame rate, compression rate, bitrate, etc.) to optimize system performance under varying resource availability while meeting latency and accuracy requirements.
Cloud-Edge-End Collaboration Mechanism: Given the limited resources of edge devices and the computational and storage-intensive nature of video inference tasks (such as video analysis and enhancement), I investigate task-layered architectures based on the complexity and timeliness of video tasks. This involves intelligently distributing tasks across the cloud, edge, and end devices. For instance, edge devices can utilize their CPUs for video preprocessing (such as frame filtering, ROI extraction, result caching, etc.), edge servers perform the actual video inference, while the cloud dynamically trains and fine-tunes models based on video content and lightweight models through pruning and compression techniques.
Collaborative Inference with Heterogeneous Devices: Single-device inference cannot guarantee real-time performance. With the widespread deployment of devices like smart cameras, I explore collaborative inference among multiple heterogeneous devices to accelerate inference. This collaboration can take two forms: (1) Model partitioning and distributed deployment: Based on the runtime status of heterogeneous devices and the size of intermediate model data, the model is divided into multiple parts and deployed across several heterogeneous devices. This pipelined approach increases throughput and reduces average processing latency. (2) Data partitioning and parallel offloading: Data is partitioned based on video features and offloaded to heterogeneous devices for parallel inference, thus accelerating the inference process.
RECRUITING
I am looking for well motivated students to work on cutting-edge research projects. Both undergraduate and graduate students are welcome!
2025年我组有空缺硕士指标,欢迎同学们联系!
NEWS












SELECTED PUBLICATIONS
ViChaser: Chase Your Viewpoint for Live Video Streaming with Block-Oriented Super Resolution
Accepted by IEEE/ACM TON 2023, CCF A link
Ning Chen, Sheng Zhang, Zhi Ma, Yu Chen, Yibo Jin, Jie Wu, Zhuzhong Qian, Yu Liang, and Sanglu Lu.
TileSR: Accelerate On-Device Super-Resolution with Parallel Offloading in Tile Granularity
Accepted by IEEE INFOCOM 2024, CCF‑A link slides
Ning Chen, Sheng Zhang, Yu Liang, Jie Wu, Yu Chen, Yuting Yan, Zhuzhong Qian and Sanglu Lu.
ResMap: Exploiting Sparse Residual Feature Map for Accelerating Cross‑Edge Video Analytics
Accepted by IEEE INFOCOM 2023, CCF‑A link slides code
Ning Chen, Shuai Zhang, Sheng Zhang, Yuting Yan, Yu Chen and Sanglu Lu.
Cuttlefish: Neural Configuration Adaptation for Video Analysis in live Augmented Reality
Accepted by IEEE TPDS 2021, CCF-A link slides code
Ning Chen, Siyi Quan, Sheng Zhang, Zhuzhong Qian, Yibo Jin, Jie Wu, Wenzhong Li, Sanglu Lu.
CONTACT
Name: Ning Chen (陈宁)
Address: Soochow University, Science and Technology Experiment Building, 430
Email: ningc@suda.edu.cn