Pub Date : 2024-10-15DOI: 10.1109/TNSE.2024.3481061
Peng Qin;Jinghan Li;Jing Zhang;Yang Fu
With the proliferation of Internet of Things (IoT), compute-intensive and latency-critical applications continue to emerge. However, IoT devices in isolated locations have insufficient energy storage as well as computing resources and may fall outside the service range of ground communication networks. To overcome the constraints of communication coverage and terminal resource, this paper proposes a multiple Unmanned Aerial Vehicle (UAV)-assisted air-ground collaborative edge computing network model, which comprises associated UAVs, auxiliary UAVs, ground user devices (GDs), and base stations (BSs), intending to minimize the overall system energy consumption. It delves into task offloading, UAV trajectory planning and edge resource allocation, which thus is classified as a Mixed-Integer Nonlinear Programming (MINLP) problem. Worse still, the coupling of long-term task queuing delay and short-term offloading decision makes it challenging to address the original issue directly. Therefore, we employ Lyapunov optimization to transform it into two sub-problems. The first involves task offloading for GDs, trajectory optimization for associated UAVs as well as auxiliary UAVs, which is tackled using Deep Reinforcement Learning (DRL), while the second deals with task partitioning and computing resource allocation, which we address via convex optimization. Through numerical simulations, we verify that the proposed approach outperforms other benchmark methods regarding overall system energy consumption.
{"title":"Joint Task Allocation and Trajectory Optimization for Multi-UAV Collaborative Air–Ground Edge Computing","authors":"Peng Qin;Jinghan Li;Jing Zhang;Yang Fu","doi":"10.1109/TNSE.2024.3481061","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3481061","url":null,"abstract":"With the proliferation of Internet of Things (IoT), compute-intensive and latency-critical applications continue to emerge. However, IoT devices in isolated locations have insufficient energy storage as well as computing resources and may fall outside the service range of ground communication networks. To overcome the constraints of communication coverage and terminal resource, this paper proposes a multiple Unmanned Aerial Vehicle (UAV)-assisted air-ground collaborative edge computing network model, which comprises associated UAVs, auxiliary UAVs, ground user devices (GDs), and base stations (BSs), intending to minimize the overall system energy consumption. It delves into task offloading, UAV trajectory planning and edge resource allocation, which thus is classified as a Mixed-Integer Nonlinear Programming (MINLP) problem. Worse still, the coupling of long-term task queuing delay and short-term offloading decision makes it challenging to address the original issue directly. Therefore, we employ Lyapunov optimization to transform it into two sub-problems. The first involves task offloading for GDs, trajectory optimization for associated UAVs as well as auxiliary UAVs, which is tackled using Deep Reinforcement Learning (DRL), while the second deals with task partitioning and computing resource allocation, which we address via convex optimization. Through numerical simulations, we verify that the proposed approach outperforms other benchmark methods regarding overall system energy consumption.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6231-6243"},"PeriodicalIF":6.7,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679304","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nowadays, owing to the communication, computation, storage, networking, and sensing abilities, the swarm of unmanned aerial vehicles (UAV) is highly anticipated to be helpful for emergency, disaster, and military situations. Additionally, in such situations, each UAV generates local sensing data with its cameras and sensors. Data sharing in UAV swarm is an urgent need for both users and administrators. For users, they may want to access data stored on any specific UAV on demand. For administrators, they need to construct global information and situational awareness to enable many cooperative applications. This paper makes the first step to tackling this open problem with an efficient data-sharing framework called Frisbee. It first groups all UAVs as a series of cells, each of which has a head-UAV. Inside any cell, all UAVs can communicate with each other directly. Thus, for the intra-cell sharing, Frisbee designs the Dynamic Cuckoo Summary for the head-UAV to accurately index all data inside the cell. For inter-cell sharing, Frisbee designs an effective method to map both the data indices and the head-UAV into a 2-dimensional virtual plane. Based on such virtual plane, a head-UAV communication graph is formed according to the communication range of each head for both data localization and transmission. The comprehensive experiments show that Frisbee achieves 14.7% higher insert throughput, 39.1% lower response delay, and 41.4% less implementation overhead, respectively, compared to the most involved solutions of the ground network.
{"title":"Frisbee: An Efficient Data Sharing Framework for UAV Swarms","authors":"Peipei Chen;Lailong Luo;Deke Guo;Qianzhen Zhang;Xueshan Luo;Bangbang Ren;Yulong Shen","doi":"10.1109/TNSE.2024.3479695","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3479695","url":null,"abstract":"Nowadays, owing to the communication, computation, storage, networking, and sensing abilities, the swarm of unmanned aerial vehicles (UAV) is highly anticipated to be helpful for emergency, disaster, and military situations. Additionally, in such situations, each UAV generates local sensing data with its cameras and sensors. Data sharing in UAV swarm is an urgent need for both users and administrators. For users, they may want to access data stored on any specific UAV on demand. For administrators, they need to construct global information and situational awareness to enable many cooperative applications. This paper makes the first step to tackling this open problem with an efficient data-sharing framework called Frisbee. It first groups all UAVs as a series of cells, each of which has a head-UAV. Inside any cell, all UAVs can communicate with each other directly. Thus, for the intra-cell sharing, Frisbee designs the Dynamic Cuckoo Summary for the head-UAV to accurately index all data inside the cell. For inter-cell sharing, Frisbee designs an effective method to map both the data indices and the head-UAV into a 2-dimensional virtual plane. Based on such virtual plane, a head-UAV communication graph is formed according to the communication range of each head for both data localization and transmission. The comprehensive experiments show that Frisbee achieves 14.7% higher insert throughput, 39.1% lower response delay, and 41.4% less implementation overhead, respectively, compared to the most involved solutions of the ground network.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"5380-5393"},"PeriodicalIF":6.7,"publicationDate":"2024-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679328","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-10DOI: 10.1109/TNSE.2024.3478174
Jinyong Chen;Rui Zhou;Yunjie Zhang;Bin Di;Guibin Sun
This paper explores information sharing within cliques to enable flexible formation pattern control of networked agents with limited communication range, where each agent is not pre-assigned a fixed point in the pattern and is unaware of the total number of agents. To achieve this, we first present a new representation of formation patterns that enables the agents to reach a consensus on the desired pattern by negotiating formation motion and agent numbers. The problem of continuously assigning each agent a point in the desired pattern is then decomposed into small size problems in terms of $delta$ -maximal cliques, which can be solved in a distributed manner. Furthermore, a maximal clique-based formation controller is employed to ensure that the agents converge to the desired pattern while preserving the connectivity of the communication topology. Simulation results demonstrate that the pattern assembly time of seven agents using the proposed algorithm is reduced by 55.1% compared with a state-of-the-art pre-assigned method, and this improvement tends to amplify with an increasing number of agents. In addition, we conduct a physical experiment involving five robots to verify the ability of the proposed algorithm in terms of formation shape assembly, manipulation, and automatic repair.
{"title":"Connectivity-Preserving Formation Control via Clique-Based Approach Without Prior Assignment","authors":"Jinyong Chen;Rui Zhou;Yunjie Zhang;Bin Di;Guibin Sun","doi":"10.1109/TNSE.2024.3478174","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3478174","url":null,"abstract":"This paper explores information sharing within cliques to enable flexible formation pattern control of networked agents with limited communication range, where each agent is not pre-assigned a fixed point in the pattern and is unaware of the total number of agents. To achieve this, we first present a new representation of formation patterns that enables the agents to reach a consensus on the desired pattern by negotiating formation motion and agent numbers. The problem of continuously assigning each agent a point in the desired pattern is then decomposed into small size problems in terms of \u0000<inline-formula><tex-math>$delta$</tex-math></inline-formula>\u0000-maximal cliques, which can be solved in a distributed manner. Furthermore, a maximal clique-based formation controller is employed to ensure that the agents converge to the desired pattern while preserving the connectivity of the communication topology. Simulation results demonstrate that the pattern assembly time of seven agents using the proposed algorithm is reduced by 55.1% compared with a state-of-the-art pre-assigned method, and this improvement tends to amplify with an increasing number of agents. In addition, we conduct a physical experiment involving five robots to verify the ability of the proposed algorithm in terms of formation shape assembly, manipulation, and automatic repair.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"5916-5929"},"PeriodicalIF":6.7,"publicationDate":"2024-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142694652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-09DOI: 10.1109/TNSE.2024.3476968
Changzhen Zhang;Jun Yang
Satellite edge computing (SEC) can offer task computing services to ground users, particularly in areas lacking terrestrial network coverage. Nevertheless, given the limited energy of low earth orbit (LEO) satellites, they cannot be used to process numerous computational tasks. Furthermore, most existing task offloading methods are designed for homogeneous tasks, which obviously cannot meet service requirements of various computational tasks. In this work, we investigate energy-efficient collaborative offloading scheme with heterogeneous tasks for SEC to save energy and improve efficiency. Firstly, by dividing computational tasks into delay-sensitive (DS) and delay-tolerant (DT) tasks, we propose a collaborative service architecture with ground edge, satellite edge, and cloud, where specific task offloading schemes are given for both sparse and dense user scenarios to reduce the energy consumption of LEO satellites. Secondly, to reduce the delay and failure rate of DS tasks, we propose an access threshold strategy for DS tasks to control the queue length and facilitate load balancing among multiple computing platforms. Thirdly, to evaluate the proposed offloading scheme, we develop the continuous-time Markov chain (CTMC) to model the traffic load on computing platforms, and the stationary distribution is solved employing the matrix-geometric method. Finally, numerical results for SEC are presented to validate the effectiveness of the proposed offloading scheme.
{"title":"An Energy-Efficient Collaborative Offloading Scheme With Heterogeneous Tasks for Satellite Edge Computing","authors":"Changzhen Zhang;Jun Yang","doi":"10.1109/TNSE.2024.3476968","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3476968","url":null,"abstract":"Satellite edge computing (SEC) can offer task computing services to ground users, particularly in areas lacking terrestrial network coverage. Nevertheless, given the limited energy of low earth orbit (LEO) satellites, they cannot be used to process numerous computational tasks. Furthermore, most existing task offloading methods are designed for homogeneous tasks, which obviously cannot meet service requirements of various computational tasks. In this work, we investigate energy-efficient collaborative offloading scheme with heterogeneous tasks for SEC to save energy and improve efficiency. Firstly, by dividing computational tasks into delay-sensitive (DS) and delay-tolerant (DT) tasks, we propose a collaborative service architecture with ground edge, satellite edge, and cloud, where specific task offloading schemes are given for both sparse and dense user scenarios to reduce the energy consumption of LEO satellites. Secondly, to reduce the delay and failure rate of DS tasks, we propose an access threshold strategy for DS tasks to control the queue length and facilitate load balancing among multiple computing platforms. Thirdly, to evaluate the proposed offloading scheme, we develop the continuous-time Markov chain (CTMC) to model the traffic load on computing platforms, and the stationary distribution is solved employing the matrix-geometric method. Finally, numerical results for SEC are presented to validate the effectiveness of the proposed offloading scheme.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6396-6407"},"PeriodicalIF":6.7,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-09DOI: 10.1109/TNSE.2024.3477453
Juan Fang;Shen Wu;Shuaibing Lu;Ziyi Teng;Huijie Chen;Neal N. Xiong
Edge computing has emerged as a promising paradigm to meet the increasing demands of latency-sensitive and computationally intensive applications. In this context, efficient server deployment and service placement are crucial for optimizing performance and increasing platform profit. This paper investigates the problem of server deployment and service placement in a multi-user scenario, aiming to enhance the profit of Mobile Network Operators while considering constraints related to distance thresholds, resource limitations, and connectivity requirements. We demonstrate that this problem is NP-hard. To address it, we propose a two-stage method to decouple the problem. In stage I, server deployment is formulated as a combinatorial optimization problem within the framework of a Markov Decision Process (MDP). We introduce the Server Deployment with Q-learning (SDQ) algorithm to establish a relatively stable server deployment strategy. In stage II, service placement is formulated as a constrained Integer Nonlinear Programming (INLP) problem. We present the Service Placement with Interior Barrier Method (SPIB) and Tree-based Branch-and-Bound (TDB) algorithms and theoretically prove their feasibility. For scenarios where the number of users changes dynamically, we propose the Distance-and-Utilization Balance Algorithm (DUBA). Extensive experiments validate the exceptional performance of our proposed algorithms in enhancing the profit.
{"title":"Enhanced Profit-Driven Optimization for Flexible Server Deployment and Service Placement in Multi-User Mobile Edge Computing Systems","authors":"Juan Fang;Shen Wu;Shuaibing Lu;Ziyi Teng;Huijie Chen;Neal N. Xiong","doi":"10.1109/TNSE.2024.3477453","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3477453","url":null,"abstract":"Edge computing has emerged as a promising paradigm to meet the increasing demands of latency-sensitive and computationally intensive applications. In this context, efficient server deployment and service placement are crucial for optimizing performance and increasing platform profit. This paper investigates the problem of server deployment and service placement in a multi-user scenario, aiming to enhance the profit of Mobile Network Operators while considering constraints related to distance thresholds, resource limitations, and connectivity requirements. We demonstrate that this problem is NP-hard. To address it, we propose a two-stage method to decouple the problem. In stage I, server deployment is formulated as a combinatorial optimization problem within the framework of a Markov Decision Process (MDP). We introduce the Server Deployment with Q-learning (SDQ) algorithm to establish a relatively stable server deployment strategy. In stage II, service placement is formulated as a constrained Integer Nonlinear Programming (INLP) problem. We present the Service Placement with Interior Barrier Method (SPIB) and Tree-based Branch-and-Bound (TDB) algorithms and theoretically prove their feasibility. For scenarios where the number of users changes dynamically, we propose the Distance-and-Utilization Balance Algorithm (DUBA). Extensive experiments validate the exceptional performance of our proposed algorithms in enhancing the profit.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6194-6206"},"PeriodicalIF":6.7,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10713190","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679382","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wi-Fi-based localization using Received Signal Strength (RSS) is widely adopted due to its cost-effectiveness and ubiquity. However, localization accuracy of RSS-based localization degrades due to random fluctuations from shadowing and multipath fading effects. Existing fading distributions like Rayleigh, $kappa ! - ! mu$ , and $alpha$ -KMS struggle to capture all factors contributing to fading. In contrast, the $alpha ! - ! eta ! - ! kappa ! - ! mu$ distribution offers the most generalized coverage of fading in literature. However, as fading distributions become more generalized, their computational demands also increases. This results in a trade-off between localization accuracy and complexity, which is undesirable for real-time localization. In this work, we propose a novel localization strategy utilizing the $alpha ! - ! eta ! - ! kappa ! - ! mu$ distribution combined with a novel approximation method that significantly reduces computational overhead while maintaining accuracy. Our proposed strategy effectively mitigates the trade-off between localization accuracy and complexity, outperforming existing state-of-the-art (SOTA) localization techniques on simulated and real-world testbeds. The proposed strategy achieves accurate localization with a speedup of 280 times over non-approximated methods. We validate its feasibility for real-time tasks on low-compute edge device Raspberry Pi Zero W, where it demonstrates fast and accurate localization, making it suitable for real-time edge applications.
利用接收信号强度（RSS）进行的基于 Wi-Fi 的定位因其成本效益高和普遍性而被广泛采用。然而，由于阴影和多径衰落效应造成的随机波动，基于 RSS 的定位精度会下降。现有的衰落分布如瑞利衰落、$kappa !- !mu$和 $alpha$-KMS 都很难捕捉到造成衰落的所有因素。相比之下，$alpha !- !eta ！- !Kappa- !mu$ 分布提供了文献中最普遍的衰减覆盖范围。然而，随着衰减分布变得越来越普遍，其计算需求也随之增加。这就导致了定位精度和复杂性之间的权衡，这对于实时定位来说是不可取的。在这项工作中，我们提出了一种利用$alpha 的新型定位策略！- !eta ！- !kappa ！- !mu$ 分布与一种新颖的近似方法相结合，在保持精度的同时显著降低了计算开销。我们提出的策略有效地缓解了定位精度和复杂性之间的权衡，在模拟和真实世界测试平台上的表现优于现有的最先进（SOTA）定位技术。与非近似方法相比，所提出的策略实现了精确定位，速度提高了 280 倍。我们在低计算能力的边缘设备 Raspberry Pi Zero W 上验证了该策略在实时任务中的可行性，它展示了快速准确的定位，使其适用于实时边缘应用。
{"title":"A Unified $alpha ! - ! eta ! -! kappa ! - ! mu$ Fading Model Based Real-Time Localization on IoT Edge Devices","authors":"Aditya Singh;Syed Danish;Gaurav Prasad;Sudhir Kumar","doi":"10.1109/TNSE.2024.3478053","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3478053","url":null,"abstract":"Wi-Fi-based localization using Received Signal Strength (RSS) is widely adopted due to its cost-effectiveness and ubiquity. However, localization accuracy of RSS-based localization degrades due to random fluctuations from shadowing and multipath fading effects. Existing fading distributions like Rayleigh, \u0000<inline-formula><tex-math>$kappa ! - ! mu$</tex-math></inline-formula>\u0000, and \u0000<inline-formula><tex-math>$alpha$</tex-math></inline-formula>\u0000-KMS struggle to capture all factors contributing to fading. In contrast, the \u0000<inline-formula><tex-math>$alpha ! - ! eta ! - ! kappa ! - ! mu$</tex-math></inline-formula>\u0000 distribution offers the most generalized coverage of fading in literature. However, as fading distributions become more generalized, their computational demands also increases. This results in a trade-off between localization accuracy and complexity, which is undesirable for real-time localization. In this work, we propose a novel localization strategy utilizing the \u0000<inline-formula><tex-math>$alpha ! - ! eta ! - ! kappa ! - ! mu$</tex-math></inline-formula>\u0000 distribution combined with a novel approximation method that significantly reduces computational overhead while maintaining accuracy. Our proposed strategy effectively mitigates the trade-off between localization accuracy and complexity, outperforming existing state-of-the-art (SOTA) localization techniques on simulated and real-world testbeds. The proposed strategy achieves accurate localization with a speedup of 280 times over non-approximated methods. We validate its feasibility for real-time tasks on low-compute edge device Raspberry Pi Zero W, where it demonstrates fast and accurate localization, making it suitable for real-time edge applications.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6207-6218"},"PeriodicalIF":6.7,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-08DOI: 10.1109/TNSE.2024.3476168
Yunkai Wei;Zikang Wan;Yinan Xiao;Supeng Leng;Kezhi Wang;Kun Yang
Unmanned Aerial Vehicles (UAV) can provide mobile edge computing (MEC) service for resource-limited devices in Internet of Things (IoT). In such scenario, partial offloading can be used to balance the computing task between the UAV and the IoT devices for higher efficiency. However, traditional partial offloading is not suitable for training deep neural network (DNN), since DNN models cannot be portioned with a continuous ratio. In this paper, we introduce a split offloading scheme, which can flexibly split the DNN training task into two parts based on the DNN layers, and allocate them to the IoT device and UAV respectively. We present a scheme to synchronize the training and communicating period of DNN layers in the UAV and IoT device, and thus reduce the model training time. Based on this scheme, an optimization model is proposed to minimize the UAV energy consumption, which jointly optimizes the UAV trajectory, the DNN split position and the service time scheduling. We divide the problem into two subproblems and solve it with an iterative solution. Simulation results show the proposed scheme can reduce the model training time and the UAV energy consumption by up to 25% and 14.4% compared with benchmark schemes, respectively.
{"title":"Joint Split Offloading and Trajectory Scheduling for UAV-Enabled Mobile Edge Computing in IoT Network","authors":"Yunkai Wei;Zikang Wan;Yinan Xiao;Supeng Leng;Kezhi Wang;Kun Yang","doi":"10.1109/TNSE.2024.3476168","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3476168","url":null,"abstract":"Unmanned Aerial Vehicles (UAV) can provide mobile edge computing (MEC) service for resource-limited devices in Internet of Things (IoT). In such scenario, partial offloading can be used to balance the computing task between the UAV and the IoT devices for higher efficiency. However, traditional partial offloading is not suitable for training deep neural network (DNN), since DNN models cannot be portioned with a continuous ratio. In this paper, we introduce a split offloading scheme, which can flexibly split the DNN training task into two parts based on the DNN layers, and allocate them to the IoT device and UAV respectively. We present a scheme to synchronize the training and communicating period of DNN layers in the UAV and IoT device, and thus reduce the model training time. Based on this scheme, an optimization model is proposed to minimize the UAV energy consumption, which jointly optimizes the UAV trajectory, the DNN split position and the service time scheduling. We divide the problem into two subproblems and solve it with an iterative solution. Simulation results show the proposed scheme can reduce the model training time and the UAV energy consumption by up to 25% and 14.4% compared with benchmark schemes, respectively.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6180-6193"},"PeriodicalIF":6.7,"publicationDate":"2024-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679272","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-07DOI: 10.1109/TNSE.2024.3475379
Ke-Jia Chen;Yaming Ji;Wenhui Mu;Youran Qu
Many real-world networks are signed networks containing positive and negative edges. The existence of negative edges in the signed graph neural network has two consequences. One is the semantic imbalance, as the negative edges are hard to obtain though they may potentially include more useful information. The other is the structural unbalance, e.g., unbalanced triangles, an indication of incompatible relationship among nodes. This paper proposes a balancing augmentation to address the two challenges. Firstly, the utility of each negative edge is determined by calculating its occurrence in balanced structures. Secondly, the original signed graph is selectively augmented with the use of (1) an edge perturbation regulator to balance the number of positive and negative edges and to determine the ratio of perturbed edges and (2) an edge utility filter to remove the negative edges with low utility. Finally, a signed graph neural network is trained on the augmented graph. The theoretical analysis is conducted to prove the effectiveness of each module and the experiments demonstrate that the proposed method can significantly improve the performance of three backbone models in link sign prediction task, with up to 22.8% in the AUC and 19.7% in F1 scores, across five real-world datasets.
{"title":"Balancing Augmentation With Edge Utility Filter for Signed Graph Neural Networks","authors":"Ke-Jia Chen;Yaming Ji;Wenhui Mu;Youran Qu","doi":"10.1109/TNSE.2024.3475379","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3475379","url":null,"abstract":"Many real-world networks are signed networks containing positive and negative edges. The existence of negative edges in the signed graph neural network has two consequences. One is the semantic imbalance, as the negative edges are hard to obtain though they may potentially include more useful information. The other is the structural unbalance, e.g., unbalanced triangles, an indication of incompatible relationship among nodes. This paper proposes a balancing augmentation to address the two challenges. Firstly, the utility of each negative edge is determined by calculating its occurrence in balanced structures. Secondly, the original signed graph is selectively augmented with the use of (1) an edge perturbation regulator to balance the number of positive and negative edges and to determine the ratio of perturbed edges and (2) an edge utility filter to remove the negative edges with low utility. Finally, a signed graph neural network is trained on the augmented graph. The theoretical analysis is conducted to prove the effectiveness of each module and the experiments demonstrate that the proposed method can significantly improve the performance of three backbone models in link sign prediction task, with up to 22.8% in the AUC and 19.7% in F1 scores, across five real-world datasets.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"5903-5915"},"PeriodicalIF":6.7,"publicationDate":"2024-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142694667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
While significant efforts have been attempted in the design, control, and optimization of complex networks, most existing works assume the network structure is known or readily available. However, the network topology can be radically recast after an adversarial attack and may remain unknown for subsequent analysis. In this work, we propose a novel Bayesian sequential learning approach to reconstruct network connectivity adaptively: A sparse Spike and Slab prior is placed on connectivity for all edges, and the connectivity learned from reconstructed nodes will be used to select the next node and update the prior knowledge. Central to our approach is that most realistic networks are sparse, in that the connectivity degree of each node is much smaller compared to the number of nodes in the network. Sequential selection of the most informative nodes is realized via the between-node expected improvement. We corroborate this sequential Bayesian approach in connectivity recovery for a synthetic ultimatum game network and the IEEE-118 power grid system. Results indicate that only a fraction (∼50%) of the nodes need to be interrogated to reveal the network topology.
{"title":"Sparse Bayesian Learning for Sequential Inference of Network Connectivity From Small Data","authors":"Jinming Wan;Jun Kataoka;Jayanth Sivakumar;Eric Peña;Yiming Che;Hiroki Sayama;Changqing Cheng","doi":"10.1109/TNSE.2024.3471852","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3471852","url":null,"abstract":"While significant efforts have been attempted in the design, control, and optimization of complex networks, most existing works assume the network structure is known or readily available. However, the network topology can be radically recast after an adversarial attack and may remain unknown for subsequent analysis. In this work, we propose a novel Bayesian sequential learning approach to reconstruct network connectivity adaptively: A sparse Spike and Slab prior is placed on connectivity for all edges, and the connectivity learned from reconstructed nodes will be used to select the next node and update the prior knowledge. Central to our approach is that most realistic networks are sparse, in that the connectivity degree of each node is much smaller compared to the number of nodes in the network. Sequential selection of the most informative nodes is realized via the between-node expected improvement. We corroborate this sequential Bayesian approach in connectivity recovery for a synthetic ultimatum game network and the IEEE-118 power grid system. Results indicate that only a fraction (∼50%) of the nodes need to be interrogated to reveal the network topology.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"5892-5902"},"PeriodicalIF":6.7,"publicationDate":"2024-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142694660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Smart contracts have been widely used for their capability of giving blockchain a user-defined logic. In recent years, several smart contract security incidents have resulted in enormous financial losses. Therefore, it is important to detect vulnerabilities in smart contracts before deployment. Machine learning has been used recently in smart contract vulnerability detection. Unfortunately, due to the loss of information during feature extraction, the detection results are unsatisfactory. Hence, we propose a novel approach called ContractGNN, which combines a new concept of a vulnerability sub-graph (VSG) with graph neural networks (GNNs). Compared with traditional methods, checking a VSG is more accurate because the VSG removes irrelevant vertexes in the control flow graph. Furthermore, a VSG can be aggregated and simplified, thus improving the efficiency of message passing in a GNN. Based on aggregated VSGs, we design a new feature extraction method that preserves semantic information, the order of opcode, and control flows of smart contracts. Moreover, we compare a large number of GNN classification models and select the best one to implement ContractGNN. We then test ContractGNN on 48,493 real-world smart contracts, and the experimental results show that ContractGNN outperforms other smart contract vulnerability detection tools, with an average F1 score of 89.70%.
{"title":"ContractGNN: Ethereum Smart Contract Vulnerability Detection Based on Vulnerability Sub-Graphs and Graph Neural Networks","authors":"Yichen Wang;Xiangfu Zhao;Long He;Zixian Zhen;Haiyue Chen","doi":"10.1109/TNSE.2024.3470788","DOIUrl":"https://doi.org/10.1109/TNSE.2024.3470788","url":null,"abstract":"Smart contracts have been widely used for their capability of giving blockchain a user-defined logic. In recent years, several smart contract security incidents have resulted in enormous financial losses. Therefore, it is important to detect vulnerabilities in smart contracts before deployment. Machine learning has been used recently in smart contract vulnerability detection. Unfortunately, due to the loss of information during feature extraction, the detection results are unsatisfactory. Hence, we propose a novel approach called ContractGNN, which combines a new concept of a \u0000<italic>vulnerability sub-graph</i>\u0000 (VSG) with \u0000<italic>graph neural networks</i>\u0000 (GNNs). Compared with traditional methods, checking a VSG is more accurate because the VSG removes irrelevant vertexes in the control flow graph. Furthermore, a VSG can be aggregated and simplified, thus improving the efficiency of message passing in a GNN. Based on aggregated VSGs, we design a new feature extraction method that preserves semantic information, the order of opcode, and control flows of smart contracts. Moreover, we compare a large number of GNN classification models and select the best one to implement ContractGNN. We then test ContractGNN on 48,493 real-world smart contracts, and the experimental results show that ContractGNN outperforms other smart contract vulnerability detection tools, with an average F1 score of 89.70%.","PeriodicalId":54229,"journal":{"name":"IEEE Transactions on Network Science and Engineering","volume":"11 6","pages":"6382-6395"},"PeriodicalIF":6.7,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142679332","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：