Jianxiong Guo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jianxiong Guo has a total of 288 co-authors. Here are the top co-authors by number of shared publications:
- Tian Wang
Institute of Artificial Intelligence and Future Networks, Beijing Normal University, Zhuhai, China
38 shared publications - Weijia Jia
Institute of Artificial Intelligence and Future Networks, Beijing Normal University, Zhuhai, China
35 shared publications - Weili Wu
Department of Computer Science, The University of Texas at Dallas, Richardson, TX, USA
35 shared publications - Zhiqing Tang
Institute of Artificial Intelligence and Future Networks, Beijing Normal University, Zhuhai, China
29 shared publications - Xingjian Ding
College of Computer Science, Beijing University of Technology, Beijing, China
18 shared publications - Aixiang Xiao
Department of Nursing Administration, Affiliated Brain Hospital of Guangzhou Medical University, Guangzhou, China
15 shared publications
Latest Publications
- Open Access
- 3 September 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1915408
Factors associated with social support among older inpatients with schizophrenia: a cross-sectional study from a tertiary psychiatric hospital in Guangzhou, China
Background: Older inpatients with schizophrenia have lower social support levels that may affect clinical outcomes, yet the associated factors in this population remain underexplored. Objective: To explore the factors associated with social support in older inpatients with schizophrenia. Methods: A cross-sectional study was conducted among 204 inpatients with schizophrenia recruited from a tertiary psychiatric hospital in Guangzhou, China, between February 2023 and December 2024. Social support and its three dimensions were assessed using the Social Support Rating Scale (SSRS). Pearson correlation analysis and multiple linear regression were applied to identify factors associated with social support. Results: The mean SSRS total score was 30.26 ± 7.07. Multiple linear regression revealed that higher levels of insight (ITAQ: β = 0.431, 95% CI: 0.308 ~ 0.555, p < 0.001) and having a spouse ( β = 4.393, 95% CI: 2.563 ~ 6.222, p < 0.001) were positively associated with higher SSRS total scores, whereas higher suicide risk (NGASR: β = −0.507, 95% CI: −0.848 ~ −0.166, p = 0.004), greater depressive symptoms (GDS: β = −0.156, 95% CI: −0.282 ~ −0.030, p = 0.016), and longer course of schizophrenia ( β = −0.092, 95% CI: −0.154 ~ −0.029, p = 0.004) were negatively associated. The regression models explained 36.3% of the variance (adjusted R 2 = 0.323, F = 9.019, p < 0.001). Conclusion: Social support among older inpatients with schizophrenia is associated with marital status, insight level, suicide risk, depressive symptoms, and disease duration. Multidimensional interventions targeting specific support dimensions are needed to improve social support in this population.
- 3 September 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1915408
Open AccessFactors associated with social support among older inpatients with schizophrenia: a cross-sectional study from a tertiary psychiatric hospital in Guangzhou, China
Background: Older inpatients with schizophrenia have lower social support levels that may affect clinical outcomes, yet the associated factors in this population remain underexplored. Objective: To explore the factors associated with social support in older inpatients with schizophrenia. Methods: A cross-sectional study was conducted among 204 inpatients with schizophrenia recruited from a tertiary psychiatric hospital in Guangzhou, China, between February 2023 and December 2024. Social support and its three dimensions were assessed using the Social Support Rating Scale (SSRS). Pearson correlation analysis and multiple linear regression were applied to identify factors associated with social support. Results: The mean SSRS total score was 30.26 ± 7.07. Multiple linear regression revealed that higher levels of insight (ITAQ: β = 0.431, 95% CI: 0.308 ~ 0.555, p < 0.001) and having a spouse ( β = 4.393, 95% CI: 2.563 ~ 6.222, p < 0.001) were positively associated with higher SSRS total scores, whereas higher suicide risk (NGASR: β = −0.507, 95% CI: −0.848 ~ −0.166, p = 0.004), greater depressive symptoms (GDS: β = −0.156, 95% CI: −0.282 ~ −0.030, p = 0.016), and longer course of schizophrenia ( β = −0.092, 95% CI: −0.154 ~ −0.029, p = 0.004) were negatively associated. The regression models explained 36.3% of the variance (adjusted R 2 = 0.323, F = 9.019, p < 0.001). Conclusion: Social support among older inpatients with schizophrenia is associated with marital status, insight level, suicide risk, depressive symptoms, and disease duration. Multidimensional interventions targeting specific support dimensions are needed to improve social support in this population.
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (18) , 43872-43882
Efficient LLM Coserving at the Edge via Resource-Aware Cooperative Scheduling
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (18) , 43872-43882
Efficient LLM Coserving at the Edge via Resource-Aware Cooperative Scheduling
- Open Access
- 14 July 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1869272
Prevalence and associated factors of cognitive impairment among hospitalized older adults with depression: a cross-sectional study from a tertiary psychiatric hospital in Guangzhou, China
Background: Geriatric depression is frequently complicated by cognitive impairment, which is linked to impaired self-care ability, elevated dementia and suicide risks, and heavier disease burden. Its pathogenesis involves metabolic disorders, vascular injury, neurodegenerative changes, and clinical factors. Nutritional status, social support, and lifestyle remain unclear in older patients with depression. This study aims to determine the prevalence of cognitive impairment and identify factors associated with cognitive impairment in hospitalized older adults with depression. Methods: A cross-sectional study was conducted at a tertiary psychiatric hospital in China from February 2021 to March 2025. A total of 196 older patients diagnosed with depression were recruited using consecutive sampling. Cognitive impairment was evaluated using the Mini Mental State Examination (MMSE). Data collection included a general information questionnaire, Mini Nutritional Assessment Short Form (MNA-SF), Brief Psychiatric Rating Scale (BPRS), Geriatric Depression Scale (GDS), Social Support Rating Scale (SSRS), and Activities of Daily Living (ADL). Multivariable logistic regression was used to identify factors associated with cognitive impairment. Results: The sample was composed of 70.9% female participants, with a mean age of 69.24 ± 5.84 years. A total of 45.4% participants had cognitive impairment. The multivariable logistic regression analysis showed that better nutritional status (MNA-SF: OR = 0.223, p = 0.010) and higher ADL score ( OR = 0.975, p = 0.021) were negatively associated with cognitive impairment, while higher BPRS score ( OR = 1.049, p = 0.020) was positively associated with cognitive impairment. Conclusion: Cognitive impairment is prevalent in hospitalized older adults with depression. Poor nutrition, low self-care ability, and severity psychiatric symptoms are associated with a higher likelihood of cognitive impairment. Targeted intervention strategies should be developed to address these factors.
- 14 July 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1869272
Open AccessPrevalence and associated factors of cognitive impairment among hospitalized older adults with depression: a cross-sectional study from a tertiary psychiatric hospital in Guangzhou, China
Background: Geriatric depression is frequently complicated by cognitive impairment, which is linked to impaired self-care ability, elevated dementia and suicide risks, and heavier disease burden. Its pathogenesis involves metabolic disorders, vascular injury, neurodegenerative changes, and clinical factors. Nutritional status, social support, and lifestyle remain unclear in older patients with depression. This study aims to determine the prevalence of cognitive impairment and identify factors associated with cognitive impairment in hospitalized older adults with depression. Methods: A cross-sectional study was conducted at a tertiary psychiatric hospital in China from February 2021 to March 2025. A total of 196 older patients diagnosed with depression were recruited using consecutive sampling. Cognitive impairment was evaluated using the Mini Mental State Examination (MMSE). Data collection included a general information questionnaire, Mini Nutritional Assessment Short Form (MNA-SF), Brief Psychiatric Rating Scale (BPRS), Geriatric Depression Scale (GDS), Social Support Rating Scale (SSRS), and Activities of Daily Living (ADL). Multivariable logistic regression was used to identify factors associated with cognitive impairment. Results: The sample was composed of 70.9% female participants, with a mean age of 69.24 ± 5.84 years. A total of 45.4% participants had cognitive impairment. The multivariable logistic regression analysis showed that better nutritional status (MNA-SF: OR = 0.223, p = 0.010) and higher ADL score ( OR = 0.975, p = 0.021) were negatively associated with cognitive impairment, while higher BPRS score ( OR = 1.049, p = 0.020) was positively associated with cognitive impairment. Conclusion: Cognitive impairment is prevalent in hospitalized older adults with depression. Poor nutrition, low self-care ability, and severity psychiatric symptoms are associated with a higher likelihood of cognitive impairment. Targeted intervention strategies should be developed to address these factors.
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 37 (7) , 1657-1670
Adaptive Image Batching and Slicing in Edge Networks for Delay-Critical Small Object Detection
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Parallel and Distributed Systems
- Vol. 37 (7) , 1657-1670
Adaptive Image Batching and Slicing in Edge Networks for Delay-Critical Small Object Detection
- Open Access
- 26 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (6) , 1-32
Stackelberg Game with Zero-Determinant Strategy for Incentive Mechanism Design in Socially Aware Mobile Crowdsensing
In Mobile Crowdsensing (MCS), incentive mechanisms are crucial for encouraging mobile users to join tasks while users selfishly pursue personal benefit maximization. While most existing studies focus on the interaction between the requester and users, the internal value of socially aware user relationships remains underexplored. Users naturally form social connections, assisting or collaborating on tasks, but current mechanisms often neglect asymmetric social effects, which can lead to unequal willingness to cooperate and eventual breakdowns in collaboration (e.g., less profitable users refusing to cooperate). To end this, we propose an integrated incentive mechanism that models the interaction between the requester and users as a two-stage Stackelberg game (SG) while accounting for pairwise asymmetric social effects. Pairwise cooperation is governed by the Iterated Prisoner's Dilemma (IPD), with users employing zero-determinant (ZD) strategies to ensure cooperation despite unequal payoffs. Additionally, a plug-and-play sub-algorithm is introduced to filter low-quality or malicious users simultaneously and evaluate task redundancy, enhancing system robustness. We rigorously prove the existence of the Nash equilibrium, design the efficient iterative algorithm for our proposed mechanism, and validate its effectiveness through extensive experiments on real-world social datasets, which demonstrate that our method significantly improves system utility and cooperation stability while ensuring quality of service (QoS) requirements.
- 26 June 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (6) , 1-32
Open AccessStackelberg Game with Zero-Determinant Strategy for Incentive Mechanism Design in Socially Aware Mobile Crowdsensing
In Mobile Crowdsensing (MCS), incentive mechanisms are crucial for encouraging mobile users to join tasks while users selfishly pursue personal benefit maximization. While most existing studies focus on the interaction between the requester and users, the internal value of socially aware user relationships remains underexplored. Users naturally form social connections, assisting or collaborating on tasks, but current mechanisms often neglect asymmetric social effects, which can lead to unequal willingness to cooperate and eventual breakdowns in collaboration (e.g., less profitable users refusing to cooperate). To end this, we propose an integrated incentive mechanism that models the interaction between the requester and users as a two-stage Stackelberg game (SG) while accounting for pairwise asymmetric social effects. Pairwise cooperation is governed by the Iterated Prisoner's Dilemma (IPD), with users employing zero-determinant (ZD) strategies to ensure cooperation despite unequal payoffs. Additionally, a plug-and-play sub-algorithm is introduced to filter low-quality or malicious users simultaneously and evaluate task redundancy, enhancing system robustness. We rigorously prove the existence of the Nash equilibrium, design the efficient iterative algorithm for our proposed mechanism, and validate its effectiveness through extensive experiments on real-world social datasets, which demonstrate that our method significantly improves system utility and cooperation stability while ensuring quality of service (QoS) requirements.
- 26 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. PP (15361233) , 1-13
Efficient Layer-Granularity Unloading for LLMs in Edge Computing
Advancements in edge computing and container technology have made it increasingly popular and convenient to deploy Large Language Models (LLMs) through containers at the edge. However, the limited GPU resources of edge servers make it impractical to retain the model in GPU memory for long periods due to the high memory cost, especially when they remain idle without user requests. Existing work unloads the entire idle models to reduce memory costs on edge servers, but reloading them introduces significant loading delays that affect task Quality of Service (QoS). Therefore, efficient management of idle models is a critical issue that has been largely neglected in existing research and requires urgent attention. To address this gap, this paper studies the problem of idle model management from the perspective of the trade-off between memory cost and loading delay under the QoS constraint. A novel layer-granularity model unloading method is proposed, which leverages the layered characteristics of the model. We formulate an online joint optimization problem to determine which layers to unload and when, and present a layer-granularity unloading strategy inspired by the ski rental problem to solve it. We implement a real system with layer-granularity unloading for LLMs on NVIDIA GPUs and validate the effectiveness of the proposed method. Experimental results show it effectively trades off memory cost and loading delay, improving overall performance by up to 39.6%.
- 26 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Mobile Computing
- Vol. PP (15361233) , 1-13
Efficient Layer-Granularity Unloading for LLMs in Edge Computing
Advancements in edge computing and container technology have made it increasingly popular and convenient to deploy Large Language Models (LLMs) through containers at the edge. However, the limited GPU resources of edge servers make it impractical to retain the model in GPU memory for long periods due to the high memory cost, especially when they remain idle without user requests. Existing work unloads the entire idle models to reduce memory costs on edge servers, but reloading them introduces significant loading delays that affect task Quality of Service (QoS). Therefore, efficient management of idle models is a critical issue that has been largely neglected in existing research and requires urgent attention. To address this gap, this paper studies the problem of idle model management from the perspective of the trade-off between memory cost and loading delay under the QoS constraint. A novel layer-granularity model unloading method is proposed, which leverages the layered characteristics of the model. We formulate an online joint optimization problem to determine which layers to unload and when, and present a layer-granularity unloading strategy inspired by the ski rental problem to solve it. We implement a real system with layer-granularity unloading for LLMs on NVIDIA GPUs and validate the effectiveness of the proposed method. Experimental results show it effectively trades off memory cost and loading delay, improving overall performance by up to 39.6%.
- Open Access
- 24 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Serum proteomic signatures and an immune-calcium signaling cascade associated with antipsychotic-related erectile dysfunction in schizophrenia
Antipsychotic-related erectile dysfunction (APRED) may represent a clinically significant yet poorly understood biological vulnerability in a subset of male patients with schizophrenia, rather than a simple pharmacological side effect. Identifying molecular mechanisms associated with APRED may facilitate early identification of susceptible individuals and inform precision treatment strategies. In a case-control design, 66 male patients with schizophrenia receiving risperidone monotherapy were stratified into APRED and non-APRED groups based on nocturnal penile tumescence monitoring. Label-free quantitative serum proteomics was performed to identify differentially expressed proteins (DEPs). Multivariate statistical analyses, functional enrichment, and protein-protein interaction (PPI) network analyses were conducted to characterize the molecular landscape associated with APRED. A total of 473 DEPs were identified between the APRED and non-APRED groups, with principal component analysis demonstrating clear group separation. Functional annotation showed significant enrichment of pathways related to immune and inflammatory responses, calcium signaling, and extracellular matrix organization. PPI network analysis of the top 40 DEPs identified 10 hub proteins, including RYR3, PRKCA, and KRT18, with RYR3 emerging as a central regulatory node linking calcium signaling and immune-related pathways. These findings identify a distinct serum proteomic signature associated with APRED in schizophrenia. The results are consistent with a hypothesized cascade characterized by gene expression reprogramming, immune and calcium signaling disruption, and subsequent structural-functional imbalance. Collectively, this study suggests that APRED may represent a peripheral manifestation of a schizophrenia-related biological subtype with intrinsic vulnerability along an immune-calcium signaling axis, providing a potential framework for early risk stratification and precision psychopharmacological management.
- 24 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessSerum proteomic signatures and an immune-calcium signaling cascade associated with antipsychotic-related erectile dysfunction in schizophrenia
Antipsychotic-related erectile dysfunction (APRED) may represent a clinically significant yet poorly understood biological vulnerability in a subset of male patients with schizophrenia, rather than a simple pharmacological side effect. Identifying molecular mechanisms associated with APRED may facilitate early identification of susceptible individuals and inform precision treatment strategies. In a case-control design, 66 male patients with schizophrenia receiving risperidone monotherapy were stratified into APRED and non-APRED groups based on nocturnal penile tumescence monitoring. Label-free quantitative serum proteomics was performed to identify differentially expressed proteins (DEPs). Multivariate statistical analyses, functional enrichment, and protein-protein interaction (PPI) network analyses were conducted to characterize the molecular landscape associated with APRED. A total of 473 DEPs were identified between the APRED and non-APRED groups, with principal component analysis demonstrating clear group separation. Functional annotation showed significant enrichment of pathways related to immune and inflammatory responses, calcium signaling, and extracellular matrix organization. PPI network analysis of the top 40 DEPs identified 10 hub proteins, including RYR3, PRKCA, and KRT18, with RYR3 emerging as a central regulatory node linking calcium signaling and immune-related pathways. These findings identify a distinct serum proteomic signature associated with APRED in schizophrenia. The results are consistent with a hypothesized cascade characterized by gene expression reprogramming, immune and calcium signaling disruption, and subsequent structural-functional imbalance. Collectively, this study suggests that APRED may represent a peripheral manifestation of a schizophrenia-related biological subtype with intrinsic vulnerability along an immune-calcium signaling axis, providing a potential framework for early risk stratification and precision psychopharmacological management.
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP (4) , 1-19
Nonlinear Group Influence Maximization Based on Prioritized Double Deep Q-Networks
Influence maximization (IM) is a fundamental problem in complex network analysis, aiming to select a subset of seed nodes subject to a budget constraint to maximize information spread. Group influence refers to the phenomenon in which individual behavior is affected by the overall influence of one’s social circle. In many real-world scenarios, such as academic collaboration networks and interest groups, collective behavior can significantly enhance information dissemination. Hypergraphs provide a natural representation for such complex relationships by modeling multinode interactions. However, existing hypergraph-based IM methods face challenges, such as the nonlinear growth of group influence and the lack of submodularity. To address these challenges, we propose a novel framework, called HIM-EDRL, to solve the Group IM problem in hypergraphs. HIM-EDRL integrates hypergraph convolution techniques to extract high-order features and frames the IM problem as a continuous-parameter optimization task using double deep Q-networks (DDQN). By incorporating group influence into the linear threshold (LT) model, our framework effectively captures the nonlinear dynamics of information spread. Furthermore, we enhance the optimization process by combining the global exploration capabilities of evolutionary algorithms (EA) with the local optimization strengths of deep reinforcement learning (DRL), overcoming limitations such as local optima and inefficient gradient-based learning. Additionally, prioritized experience replay (PER) is employed to improve training efficiency by focusing on critical experiences. Experiments on seven real-world hypergraphs demonstrate consistent gains over competitive baselines. We further analyze time and memory complexity and discuss practical scalability limits.
- 22 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP (4) , 1-19
Nonlinear Group Influence Maximization Based on Prioritized Double Deep Q-Networks
Influence maximization (IM) is a fundamental problem in complex network analysis, aiming to select a subset of seed nodes subject to a budget constraint to maximize information spread. Group influence refers to the phenomenon in which individual behavior is affected by the overall influence of one’s social circle. In many real-world scenarios, such as academic collaboration networks and interest groups, collective behavior can significantly enhance information dissemination. Hypergraphs provide a natural representation for such complex relationships by modeling multinode interactions. However, existing hypergraph-based IM methods face challenges, such as the nonlinear growth of group influence and the lack of submodularity. To address these challenges, we propose a novel framework, called HIM-EDRL, to solve the Group IM problem in hypergraphs. HIM-EDRL integrates hypergraph convolution techniques to extract high-order features and frames the IM problem as a continuous-parameter optimization task using double deep Q-networks (DDQN). By incorporating group influence into the linear threshold (LT) model, our framework effectively captures the nonlinear dynamics of information spread. Furthermore, we enhance the optimization process by combining the global exploration capabilities of evolutionary algorithms (EA) with the local optimization strengths of deep reinforcement learning (DRL), overcoming limitations such as local optima and inefficient gradient-based learning. Additionally, prioritized experience replay (PER) is employed to improve training efficiency by focusing on critical experiences. Experiments on seven real-world hypergraphs demonstrate consistent gains over competitive baselines. We further analyze time and memory complexity and discuss practical scalability limits.
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 5271-5286
Fine-Grained Lifetime Control for Heterogeneous Service Provisioning in Energy-Constrained Edge-Edge Systems
To support delay-critical applications, migrating services from the cloud to edge servers (ESs) can effectively reduce service delay. However, in such a resource-constrained scenario, providing heterogeneous services to meet diverse user needs poses significant challenges. Specifically, ESs have limited computational resources and are often energy-constrained, complicating service placement and provisioning. Existing studies propose collaboration schemes to improve resource utilization and reduce service delay. Nevertheless, these works heavily depend on full-service coverage by the cloud or rely on coarse-grained (e.g., time-cycle level) strategies that inevitably waste energy in idle time slots, which are unsuitable for energy-constrained settings and dynamic edge environments. To address these challenges, we propose a novel edge-edge collaboration method tailored for heterogeneous edge service provisioning in energy-constrained networks. First, we formulate the heterogeneous service provisioning problem with the objective of delay minimization under energy constraints and prove its NP-hardness. Our framework leverages latest task statistics to decide the service lifetime in a fine-grained time slot level, so that enables edge collaboration to maximize resource utilization and adaptability in resource-limited conditions. Specifically, we decompose the problem into three subproblems: service placement, service lifetime decision, and task scheduling, and we design targeted lightweight solutions for each, ensuring low delay and efficient energy usage. Finally, we validate our method through comprehensive simulations on real-world datasets and implement it on a testbed with three ESs. Results demonstrate that our approach reduces service delays by an average of 69.6% across various energy-constrained scenarios.
- 8 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 5271-5286
Fine-Grained Lifetime Control for Heterogeneous Service Provisioning in Energy-Constrained Edge-Edge Systems
To support delay-critical applications, migrating services from the cloud to edge servers (ESs) can effectively reduce service delay. However, in such a resource-constrained scenario, providing heterogeneous services to meet diverse user needs poses significant challenges. Specifically, ESs have limited computational resources and are often energy-constrained, complicating service placement and provisioning. Existing studies propose collaboration schemes to improve resource utilization and reduce service delay. Nevertheless, these works heavily depend on full-service coverage by the cloud or rely on coarse-grained (e.g., time-cycle level) strategies that inevitably waste energy in idle time slots, which are unsuitable for energy-constrained settings and dynamic edge environments. To address these challenges, we propose a novel edge-edge collaboration method tailored for heterogeneous edge service provisioning in energy-constrained networks. First, we formulate the heterogeneous service provisioning problem with the objective of delay minimization under energy constraints and prove its NP-hardness. Our framework leverages latest task statistics to decide the service lifetime in a fine-grained time slot level, so that enables edge collaboration to maximize resource utilization and adaptability in resource-limited conditions. Specifically, we decompose the problem into three subproblems: service placement, service lifetime decision, and task scheduling, and we design targeted lightweight solutions for each, ensuring low delay and efficient energy usage. Finally, we validate our method through comprehensive simulations on real-world datasets and implement it on a testbed with three ESs. Results demonstrate that our approach reduces service delays by an average of 69.6% across various energy-constrained scenarios.
- Open Access
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 5239-5254
Decentralized Online Learning With Hard Constraints: A Doubly-Bounded Queue Approach
We consider decentralized online convex optimization with time-varying constraints and conduct performance analysis using two stringent metrics: network dynamic regret with respect to the online global solution benchmark, and hard constraint violation that does not allow any compensated violation over time and learners. We propose an efficient algorithm called Constrained Online Learning with Doubly-bounded Queue (COLDQ), which introduces a novel virtual queue that is both lower and upper bounded, allowing tight control of the constraint violation without needing the Slater’s condition. We prove via a new Lyapunov drift analysis that COLDQ provides O(T 1+Vx/2 ) network dynamic regret and O(min{T 3+Vx/4 ; TVg }) hard constraint violation, where Vx and Vg capture the dynamics of the loss and constraint functions. For the first time, the two bounds smoothly approach to the best-known O(T 1/2 ) regret and O(1) violation, as the dynamics of the losses and constraints diminish, under both centralized and decentralized settings. For strongly convex loss functions, COLDQ provides O(log T) static regret and O(min{T1/2 (log T)1/2 , TVg }) hard constraint violation. Simulation results based on synthetic and canonical datasets demonstrate that COLDQ substantially outperforms the state-of-the-art approaches for both convex and non-convex loss functions.
- 6 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Networking
- Vol. 34, 5239-5254
Open AccessDecentralized Online Learning With Hard Constraints: A Doubly-Bounded Queue Approach
We consider decentralized online convex optimization with time-varying constraints and conduct performance analysis using two stringent metrics: network dynamic regret with respect to the online global solution benchmark, and hard constraint violation that does not allow any compensated violation over time and learners. We propose an efficient algorithm called Constrained Online Learning with Doubly-bounded Queue (COLDQ), which introduces a novel virtual queue that is both lower and upper bounded, allowing tight control of the constraint violation without needing the Slater’s condition. We prove via a new Lyapunov drift analysis that COLDQ provides O(T 1+Vx/2 ) network dynamic regret and O(min{T 3+Vx/4 ; TVg }) hard constraint violation, where Vx and Vg capture the dynamics of the loss and constraint functions. For the first time, the two bounds smoothly approach to the best-known O(T 1/2 ) regret and O(1) violation, as the dynamics of the losses and constraints diminish, under both centralized and decentralized settings. For strongly convex loss functions, COLDQ provides O(log T) static regret and O(min{T1/2 (log T)1/2 , TVg }) hard constraint violation. Simulation results based on synthetic and canonical datasets demonstrate that COLDQ substantially outperforms the state-of-the-art approaches for both convex and non-convex loss functions.