Zihan Fang

Zihan Fang

Auto-generated profile, based on publicly available information

Zihan Fang

City University of Hong Kong

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 October 2024 in IEEE Transactions on Mobile Computing
Verified
Journal Article
Closed access
ZLZheng Lin
Zheng Lin
ZCZhe Chen
Zhe Chen
ZFZihan Fang
Zihan Fang
Xianhao ChenXianhao Chen
Xianhao Chen
XWXiong Wang
Xiong Wang
Yue GaoYue Gao
Yue Gao
Recently, a large number of Low Earth Orbit (LEO) satellites have been launched and deployed successfully in space. Due to multimodal sensors equipped by the LEO satellites, they serve not only for communication but also for various machine learning applications. However, the ground station (GS) may be incapable of downloading such a large volume of raw sensing data for centralized model training due to the limited contact time with LEO satellites (e.g. 5 minutes). Therefore, federated learning (FL) has emerged as the promising solution to address this problem via on-device training. Unfortunately, enabling FL on LEO satellites still face three critical challenges: i) heterogeneous computing and memory capabilities, ii) limited downlink/uplink rate, and iii) model staleness. To this end, we propose FedSN as a general FL framework to tackle the above challenges. Specifically, we first present a novel sub-structure scheme to enable heterogeneous local model training considering different computing, memory, and communication constraints on LEO satellites. Additionally, we propose a pseudo-synchronous model aggregation strategy to dynamically schedule model aggregation for compensating model staleness. Extensive experiments with real-world satellite data demonstrate that FedSN framework achieves higher accuracy, lower computing, and communication overhead than the state-of-the-art benchmarks.

31 December 2025 in IEEE Transactions on Mobile Computing
Verified
Journal Article
Closed access
ZFZihan Fang
Zihan Fang
ZLZheng Lin
Zheng Lin
ZCZhe Chen
Zhe Chen
Xianhao ChenXianhao Chen
Xianhao Chen
Yue GaoYue Gao
Yue Gao
YFYuguang Fang
Yuguang Fang
Recently, there has been a surge in the development of advanced intelligent generative content (AIGC), especially large language models (LLMs). For many downstream tasks, it is necessary to fine-tune LLMs using private data. While federated learning offers a promising privacy-preserving solution to LLM fine-tuning, the substantial size of an LLM, combined with high computational and communication demands, makes it hard to apply to handle downstream tasks. More importantly, private edge servers often possess varying computing and network resources in real-world scenarios, introducing additional complexities to LLM fine-tuning. To tackle these problems, we design and implement an automated federated pipeline, named FedPipe, to fine-tune LLMs on heterogeneous edge servers with minimal training cost and no additional inference latency. FedPipe firstly identifies the weights to be fine-tuned based on their contributions to the LLM training. It then configures a low-rank adapter for each selected weight within the resource constraints of the edge server and aggregates these local adapters from all edge servers to fine-tune the whole LLM. Finally, it appropriately quantizes the parameters of LLM to reduce memory consumption according to the requirements of edge servers. Extensive experiments demonstrate that FedPipe expedites model training and achieves higher accuracy than the state-of-the-art benchmarks.

8 August 2020 in Sensors
Verified
Journal Article
Open access
PDF file
QYQin Yang
Qin Yang
PWPeizhi Wang
Peizhi Wang
ZFZihan Fang
Zihan Fang
QLQiyong Lu
Qiyong Lu
The occlusion problem is very common in pedestrian retrieval scenarios. When persons are occluded by various obstacles, the noise caused by the occluded area greatly affects the retrieval results. However, many previous pedestrian re-identification (Re-ID) methods ignore this problem. To solve it, we propose a semantic-guided alignment model that uses image semantic information to separate useful information from occlusion noise. In the image preprocessing phase, we use a human semantic parsing network to generate probability maps. These maps show which regions of images are occluded, and the model automatically crops images to preserve the visible parts. In the construction phase, we fuse the probability maps with the global features of the image, and semantic information guides the model to focus on visible human regions and extract local features. During the matching process, we propose a measurement strategy that only calculates the distance of public areas (visible human areas on both images) between images, thereby suppressing the spatial misalignment caused by non-public areas. Experimental results on a series of public datasets confirm that our method outperforms previous occluded Re-ID methods, and it achieves top performance in the holistic Re-ID problem.

3 April 2026 in IEEE Transactions on Mobile Computing
Verified
Journal Article
Closed access
ZLZheng Lin
Zheng Lin
YZYuxin Zhang
Yuxin Zhang
ZCZhe Chen
Zhe Chen
ZFZihan Fang
Zihan Fang
Xianhao ChenXianhao Chen
Xianhao Chen
Yue GaoYue Gao
Yue Gao
Recently, large language models (LLMs) have achieved remarkable breakthroughs, revolutionizing the natural language processing domain and beyond. Due to immense parameter sizes, fine-tuning these models with private data for diverse downstream tasks has become mainstream. Though federated learning (FL) offers a promising solution for fine-tuning LLMs without sharing raw data, substantial computing costs hinder its democratization. Moreover, in real-world scenarios, private client devices often possess heterogeneous computing resources, further complicating LLM fine-tuning. To combat these challenges, we propose HSplitLoRA, a heterogeneous parameter-efficient fine-tuning (PEFT) framework built on split learning (SL) and low-rank adaptation (LoRA) fine-tuning, for efficiently fine-tuning LLMs on heterogeneous client devices. HSplitLoRA first identifies important weights based on their contributions to LLM training. It then dynamically configures the decomposition ranks of LoRA adapters for selected weights and determines the model split point according to varying computing budgets of client devices. Finally, a noise-free adapter aggregation mechanism is devised to support heterogeneous adapter aggregation without introducing noise. Extensive experiments demonstrate that HSplitLoRA outperforms state-of-the-art benchmarks in training accuracy and convergence speed.

21 August 2025 in IEEE Transactions on Mobile Computing
Verified
Journal Article
Closed access
ZLZheng Lin
Zheng Lin
YZYuxin Zhang
Yuxin Zhang
ZCZhe Chen
Zhe Chen
ZFZihan Fang
Zihan Fang
CWCong Wu
Cong Wu
Jun LuoJun Luo
Jun Luo
Recently, the increasing deployment of LEO satellite systems has enabled various space analytics (e.g., crop and climate monitoring), which heavily relies on the advancements in deep learning (DL). However, the intermittent connectivity between LEO satellites and ground station (GS) significantly hinders the timely transmission of raw data to GS for centralized learning, while the scaled-up DL models hamper distributed learning on resource-constrained LEO satellites. Though split learning (SL) can be a potential solution to these problems by partitioning a model and offloading primary training workload to GS, the labor-intensive labeling process remains an obstacle, with intermittent connectivity and data heterogeneity being other challenges. In this paper, we propose LEO-Split, a semi-supervised (SS) SL design tailored for satellite networks to combat these challenges. Leveraging SS learning to handle (labeled) data scarcity, we construct an auxiliary model to tackle the training failure of the satellite-GS non-contact time. Moreover, we propose a pseudo-labeling algorithm to rectify data imbalances across satellites. Lastly, an adaptive activation interpolation scheme is devised to prevent the overfitting of server-side sub-model training at GS. Extensive experiments with real-world LEO satellite traces (e.g., Starlink) demonstrate that our LEO-Split framework achieves superior performance compared to state-of-the-art benchmarks.

30 June 2025 in IEEE Transactions on Cognitive Communications and Networking
Verified
Journal Article
Closed access
HYHaoxuan Yuan
Haoxuan Yuan
ZCZhe Chen
Zhe Chen
ZLZheng Lin
Zheng Lin
JPJinbo Peng
Jinbo Peng
ZFZihan Fang
Zihan Fang
Yue GaoYue Gao
Yue Gao
Low Earth Orbit (LEO) satellite Internet, providing global connectivity through non-terrestrial networks, has intensified pressure on limited spectrum resources due to the large-scale deployment of both non-terrestrial and terrestrial networks. Consequently, dynamic spectrum sharing is essential for coexistence, with accurate spectrum sensing being critical. However, single-satellite spectrum sensing is prone to instability from fluctuating channel conditions, which can severely degrade accuracy. To address this, we propose a multi-satellite collaborative sensing approach. Collaborative spectrum sensing faces challenges such as channel heterogeneity among satellites, limited data transmission capacity between satellites and ground stations, and the impact of packet loss on transmitted data. To overcome these issues, we model the sensing data as a graph and leverage a graph learning-based algorithm to enhance sensing accuracy. Additionally, we adopt a hybrid data compression strategy that combines sub-Nyquist sampling with an autoencoder to significantly reduce the volume of transmitted data. A contrastive learning-based mechanism is also integrated to mitigate the effects of packet loss. Extensive experiments demonstrate that our multi-satellite approach improves spectrum sensing performance under unstable channel conditions. Even in low signal-to-noise ratio conditions, the method remains robust. Furthermore, our graph learning technique outperforms traditional deep learning methods in terms of accuracy.