Zongpeng Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Zongpeng Li has a total of 241 co-authors. Here are the top co-authors by number of shared publications:
- Qifu Tyler Sun
School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China
12 shared publications - Chuan Wu
The University of Hong Kong, Hong Kong, China
11 shared publications - Ruiting Zhou
Southeast University
10 shared publications Chuan Wu
Department of Computer Science The University of Hong Kong
6 shared publicationsLina Su
School of Computer Science, Wuhan University, Wuhan, China
6 shared publications- Xianglong Li
School of Computer Science, Wuhan University, Wuhan, China
6 shared publications
Latest Publications
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 25 (12) , 22745-22761
Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. 25 (12) , 22745-22761
Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services
- 7 October 2026
- Published by Springer Nature
- p. 275-296
Detecting Adversarial Illicit Promotional Images via Threat-Driven Multimodal Analysis
- 7 October 2026
- Published by Springer Nature
- p. 275-296
Detecting Adversarial Illicit Promotional Images via Threat-Driven Multimodal Analysis
- 30 July 2026
- Published by Elsevier in Computer Networks
- Vol. 288
PADA: An online scheduling framework for UAV emergency logistics in dynamic disaster environments
- 30 July 2026
- Published by Elsevier in Computer Networks
- Vol. 288
PADA: An online scheduling framework for UAV emergency logistics in dynamic disaster environments
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Targeted Bit-flip Attacks on LLM-Based Agents
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessTargeted Bit-flip Attacks on LLM-Based Agents
- Open Access
- 21 July 2026
- Published by Elsevier in SSRN Electronic Journal
Joint Online Offloading and Caching in Energy-Harvesting Edge Networks
Multi-tier mobile edge computing (MEC) faces two control problems that are almost always optimized in isolation: where to execute latency- and deadline-sensitive tasks (computation offloading across end devices, edge servers, and the cloud) and what to cache at the network edge (service/content caching). These decisions are fundamentally coupled: a cache hit makes edge execution cheap and thereby changes the optimal routing, while routing demand determines what is worth caching. We study their joint control online in energy-harvesting cloud-edge-end networks, where future task arrivals, channel conditions, and harvested energy are unknown, deadlines are tight, and device batteries are finite. We propose JOCO (Joint Online Cloud-Edge-End Orchestration), a Lyapunov drift-plus-penalty controller that makes routing cache-aware, steers caching toward the offloading demand it actually serves, and tracks finite energy-harvesting batteries through virtual queues, all at near-linear complexity per slot. JOCO attains the standard [O(1/V), O(V)] cost-versus-backlog tradeoff and guarantees per-slot integral feasibility online. On an event-driven simulator, JOCO achieves the lowest total cost on every random seed and the lowest deadline-violation rate among cost-competitive methods, cutting violations by about 50% over a cache-unaware Lyapunov baseline while scaling to N=1000 devices at about 10 ms per slot.
- 21 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessJoint Online Offloading and Caching in Energy-Harvesting Edge Networks
Multi-tier mobile edge computing (MEC) faces two control problems that are almost always optimized in isolation: where to execute latency- and deadline-sensitive tasks (computation offloading across end devices, edge servers, and the cloud) and what to cache at the network edge (service/content caching). These decisions are fundamentally coupled: a cache hit makes edge execution cheap and thereby changes the optimal routing, while routing demand determines what is worth caching. We study their joint control online in energy-harvesting cloud-edge-end networks, where future task arrivals, channel conditions, and harvested energy are unknown, deadlines are tight, and device batteries are finite. We propose JOCO (Joint Online Cloud-Edge-End Orchestration), a Lyapunov drift-plus-penalty controller that makes routing cache-aware, steers caching toward the offloading demand it actually serves, and tracks finite energy-harvesting batteries through virtual queues, all at near-linear complexity per slot. JOCO attains the standard [O(1/V), O(V)] cost-versus-backlog tradeoff and guarantees per-slot integral feasibility online. On an event-driven simulator, JOCO achieves the lowest total cost on every random seed and the lowest deadline-violation rate among cost-competitive methods, cutting violations by about 50% over a cache-unaware Lyapunov baseline while scaling to N=1000 devices at about 10 ms per slot.
- Open Access
- 22 May 2026
- Published by Elsevier in SSRN Electronic Journal
Online Multi-Slice Task Allocation in Multi-UAV Networks
UAV network slicing enables flexible, task-oriented resource allocation. However, most current deployments lack true slicing support and face scalability and coordination challenges. Existing methods fail to jointly handle multi-type tasks, UAV deadlines, and hierarchical bandwidth quotas, leading to suboptimal welfare in dynamic environments. We study QoS-aware resource allocation in multi-UAV networks, with the objective of maximizing social welfare. We present a holistic mechanism called Advanced Dynamic Slicing and Task Allocation that supports per-task cross-slice orchestration and per-UAV multi-slice service under per-UAV capacity, coverage, and deadline constraints; we design an online primal–dual algorithm with a dual subroutine that prices multi-slice demands and applies a unit-cost heuristic (travel cost amortized by multi-slice bandwidth requirement) to select feasible UAVs; and we reformulate the integer program to capture travel constraints, bidding prices, and slice-level quotas, using complementary-slackness guidance for real-time task acceptance and resource allocation. In contrast to existing online primal–dual or auction-based UAV schedulers, the proposed mechanism explicitly combines (i) per-UAV concurrent multi-slice service, (ii) per-task cross-slice composition, and (iii) travel-cost-aware deadline constraints in a single optimization and algorithmic framework. Our duality analysis establishes a one-to-one correspondence between dual solutions and primal allocations, ensuring feasible online decisions. Empirical studies verify consistent performance gains, achieving about 10\% higher social welfare than other allocation baselines. We also discuss modeling assumptions, economic interpretation, and limitations, providing a clearer positioning of the proposed mechanism relative to prior work and outlining directions for further empirical validation.
- 22 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessOnline Multi-Slice Task Allocation in Multi-UAV Networks
UAV network slicing enables flexible, task-oriented resource allocation. However, most current deployments lack true slicing support and face scalability and coordination challenges. Existing methods fail to jointly handle multi-type tasks, UAV deadlines, and hierarchical bandwidth quotas, leading to suboptimal welfare in dynamic environments. We study QoS-aware resource allocation in multi-UAV networks, with the objective of maximizing social welfare. We present a holistic mechanism called Advanced Dynamic Slicing and Task Allocation that supports per-task cross-slice orchestration and per-UAV multi-slice service under per-UAV capacity, coverage, and deadline constraints; we design an online primal–dual algorithm with a dual subroutine that prices multi-slice demands and applies a unit-cost heuristic (travel cost amortized by multi-slice bandwidth requirement) to select feasible UAVs; and we reformulate the integer program to capture travel constraints, bidding prices, and slice-level quotas, using complementary-slackness guidance for real-time task acceptance and resource allocation. In contrast to existing online primal–dual or auction-based UAV schedulers, the proposed mechanism explicitly combines (i) per-UAV concurrent multi-slice service, (ii) per-task cross-slice composition, and (iii) travel-cost-aware deadline constraints in a single optimization and algorithmic framework. Our duality analysis establishes a one-to-one correspondence between dual solutions and primal allocations, ensuring feasible online decisions. Empirical studies verify consistent performance gains, achieving about 10\% higher social welfare than other allocation baselines. We also discuss modeling assumptions, economic interpretation, and limitations, providing a clearer positioning of the proposed mechanism relative to prior work and outlining directions for further empirical validation.
- 9 April 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 323
Federated learning of diffusion networks
- 9 April 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 323
Federated learning of diffusion networks
- 18 March 2026
- Published by Elsevier in Neural Networks
- Vol. 201, 108872
Position-Sensitive painterly image harmonization
- 18 March 2026
- Published by Elsevier in Neural Networks
- Vol. 201, 108872
Position-Sensitive painterly image harmonization
- 14 March 2026
- Published by Elsevier in Computer Networks
- Vol. 281
Hermes: Multi-job federated learning with switching cost in wireless networks
- 14 March 2026
- Published by Elsevier in Computer Networks
- Vol. 281
Hermes: Multi-job federated learning with switching cost in wireless networks
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Theory
- Vol. 72 (5) , 2861-2880
Construction of MRD Codes Based on Circular-Shift Operations
Most well-known constructions of $(N \times n, q^{Nk}, d)$ maximum rank distance (MRD) codes rely on the arithmetic of $\mathbb {F}_{q^{N}}$ , whose increasing complexity with larger $N$ hinders parameter selection and practical implementation. In this work, based on circular-shift operations, we present a construction of $(J \times n, q^{Jk}, d)$ MRD codes with efficient encoding, where $J$ equals to the Euler’s totient function of a defined $L$ subject to $\gcd (q, L) = 1$ . The proposed construction is performed entirely over $\mathbf {F}_{q}$ and avoids the arithmetic of $\mathbf {F}_{q^{J}}$ . Furthermore, we characterize the constructed MRD codes, Gabidulin codes and twisted Gabidulin codes using a set of $q$ -linearized polynomials over the row vector space $\mathbb {F}_{q}^{N}$ , and clarify the inherent difference and connection among these code families. Specifically, for the case $J \neq m_{L}$ , where $m_{L}$ denotes the multiplicative order of $q$ modulo $L$ , we show that the proposed MRD codes, in a family of settings, are different from any Gabidulin code and any twisted Gabidulin code. For the case $J = m_{L}$ , we prove that every constructed $(J \times n, q^{Jk}, d)$ MRD code coincides with a $(J \times n, q^{Jk}, d)$ Gabidulin code, so that an equivalent circular-shift-based construction of Gabidulin codes is obtained which operates directly over $\mathbb {F}_{q}$ and avoids the arithmetic of $\mathbb {F}_{q^{J}}$ . In addition, we prove that under some parameter settings, the constructed MRD codes are equivalent to a generalization of Gabidulin codes obtained by summing and concatenating several $(m_{L} \times n, q^{m_{L}k}, d)$ Gabidulin codes. When $q=2$ , $L$ is prime and $n\leq m_{L}$ , it is analyzed that generating a codeword of the proposed $((L-1) \times n, 2^{(L-1)k}, d)$ MRD codes requires $O(nkL)$ exclusive OR (XOR) operations, while generating a codeword of $((L-1) \times n, 2^{(L-1)k}, d)$ Gabidulin codes, based on customary construction, requires $O(nkL^{2})$ XOR operations.
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Information Theory
- Vol. 72 (5) , 2861-2880
Construction of MRD Codes Based on Circular-Shift Operations
Most well-known constructions of $(N \times n, q^{Nk}, d)$ maximum rank distance (MRD) codes rely on the arithmetic of $\mathbb {F}_{q^{N}}$ , whose increasing complexity with larger $N$ hinders parameter selection and practical implementation. In this work, based on circular-shift operations, we present a construction of $(J \times n, q^{Jk}, d)$ MRD codes with efficient encoding, where $J$ equals to the Euler’s totient function of a defined $L$ subject to $\gcd (q, L) = 1$ . The proposed construction is performed entirely over $\mathbf {F}_{q}$ and avoids the arithmetic of $\mathbf {F}_{q^{J}}$ . Furthermore, we characterize the constructed MRD codes, Gabidulin codes and twisted Gabidulin codes using a set of $q$ -linearized polynomials over the row vector space $\mathbb {F}_{q}^{N}$ , and clarify the inherent difference and connection among these code families. Specifically, for the case $J \neq m_{L}$ , where $m_{L}$ denotes the multiplicative order of $q$ modulo $L$ , we show that the proposed MRD codes, in a family of settings, are different from any Gabidulin code and any twisted Gabidulin code. For the case $J = m_{L}$ , we prove that every constructed $(J \times n, q^{Jk}, d)$ MRD code coincides with a $(J \times n, q^{Jk}, d)$ Gabidulin code, so that an equivalent circular-shift-based construction of Gabidulin codes is obtained which operates directly over $\mathbb {F}_{q}$ and avoids the arithmetic of $\mathbb {F}_{q^{J}}$ . In addition, we prove that under some parameter settings, the constructed MRD codes are equivalent to a generalization of Gabidulin codes obtained by summing and concatenating several $(m_{L} \times n, q^{m_{L}k}, d)$ Gabidulin codes. When $q=2$ , $L$ is prime and $n\leq m_{L}$ , it is analyzed that generating a codeword of the proposed $((L-1) \times n, 2^{(L-1)k}, d)$ MRD codes requires $O(nkL)$ exclusive OR (XOR) operations, while generating a codeword of $((L-1) \times n, 2^{(L-1)k}, d)$ Gabidulin codes, based on customary construction, requires $O(nkL^{2})$ XOR operations.