Pub Date : 2024-01-30DOI: 10.1109/TGCN.2024.3360242
Chang Liu;Jun-Bo Wang;Cheng Zeng;Yijian Chen;Hongkang Yu;Yijin Pan
Multi-access edge computing (MEC) and wireless power transfer (WPT) have emerged as promising paradigms to address the bottlenecks of computing power and battery capacity of mobile devices. In this paper, we investigate the integrated scheduling of WPT and task offloading in a rechargeable multi-access edge computing network (RMECN). Specifically, we focus on exploring the tradeoff between energy efficiency, buffer stability, and battery level stability in the RMECN to obtain reasonable scheduling. In addition, we adopt a dynamic Li-ion battery model to describe the charge/discharge characteristics. Given the stochastic nature of channel states and task arrivals, we formulate a stochastic optimization problem that minimizes system energy consumption while ensuring buffer and battery level stability. In this problem, we jointly consider offloading decisions, local central processing unit (CPU) frequency, transmission power, and current of charge/discharge as optimization variables. To solve this stochastic non-convex problem, we first transform it into an online optimization problem using the Lyapunov optimization theory. Then, we propose a distributed algorithm based on game theory to overcome the excessive computation and time consumption of traditional centralized optimization algorithms. The numerical results demonstrate that the proposed tradeoff scheme and corresponding algorithm can effectively reduce the system’s energy consumption while ensuring the stability of buffer and battery level.
{"title":"Joint Optimization of Transmission and Computation Resources for Rechargeable Multi-Access Edge Computing Networks","authors":"Chang Liu;Jun-Bo Wang;Cheng Zeng;Yijian Chen;Hongkang Yu;Yijin Pan","doi":"10.1109/TGCN.2024.3360242","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3360242","url":null,"abstract":"Multi-access edge computing (MEC) and wireless power transfer (WPT) have emerged as promising paradigms to address the bottlenecks of computing power and battery capacity of mobile devices. In this paper, we investigate the integrated scheduling of WPT and task offloading in a rechargeable multi-access edge computing network (RMECN). Specifically, we focus on exploring the tradeoff between energy efficiency, buffer stability, and battery level stability in the RMECN to obtain reasonable scheduling. In addition, we adopt a dynamic Li-ion battery model to describe the charge/discharge characteristics. Given the stochastic nature of channel states and task arrivals, we formulate a stochastic optimization problem that minimizes system energy consumption while ensuring buffer and battery level stability. In this problem, we jointly consider offloading decisions, local central processing unit (CPU) frequency, transmission power, and current of charge/discharge as optimization variables. To solve this stochastic non-convex problem, we first transform it into an online optimization problem using the Lyapunov optimization theory. Then, we propose a distributed algorithm based on game theory to overcome the excessive computation and time consumption of traditional centralized optimization algorithms. The numerical results demonstrate that the proposed tradeoff scheme and corresponding algorithm can effectively reduce the system’s energy consumption while ensuring the stability of buffer and battery level.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 3","pages":"1259-1272"},"PeriodicalIF":5.3,"publicationDate":"2024-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142123029","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 : 2024-01-29DOI: 10.1109/TGCN.2024.3359208
Siting Lv;Xiaohui Li;Jiawen Liu;Mingli Shi
This paper focuses on a deep learning (DL) framework for the Sub-6G aided millimeter-wave (mmWave) communication system, aiming to reduce the overhead of mmWave systems. The proposed framework consists of two-stage cascaded networks, named HestNet and HBFNet, for mmWave channel estimation and hybrid beamforming (HBF) design, respectively. The number of parameters for channel estimation is reduced by using channel covariance matrix (CCM) estimation instead. However, a new challenge of estimating high-dimensional data from low-dimensional data should be considered since the dimension of Sub-6G channel data is much smaller than that of mmWave. Subsequently, a data deformation approach is introduced into the framework to match the size of Sub-6G channel data with that of mmWave. The simulation results show that the application of statistical channel information based on Sub-6G channel information to aid mmWave communication is reasonable and effective, it achieves good estimation performance and spectral efficiency. Moreover, the two-stage cascaded network architecture proposed in this paper is also more robust to channel estimation errors.
{"title":"Sub-6G Aided Millimeter Wave Hybrid Beamforming: A Two-Stage Deep Learning Framework With Statistical Channel Information","authors":"Siting Lv;Xiaohui Li;Jiawen Liu;Mingli Shi","doi":"10.1109/TGCN.2024.3359208","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3359208","url":null,"abstract":"This paper focuses on a deep learning (DL) framework for the Sub-6G aided millimeter-wave (mmWave) communication system, aiming to reduce the overhead of mmWave systems. The proposed framework consists of two-stage cascaded networks, named HestNet and HBFNet, for mmWave channel estimation and hybrid beamforming (HBF) design, respectively. The number of parameters for channel estimation is reduced by using channel covariance matrix (CCM) estimation instead. However, a new challenge of estimating high-dimensional data from low-dimensional data should be considered since the dimension of Sub-6G channel data is much smaller than that of mmWave. Subsequently, a data deformation approach is introduced into the framework to match the size of Sub-6G channel data with that of mmWave. The simulation results show that the application of statistical channel information based on Sub-6G channel information to aid mmWave communication is reasonable and effective, it achieves good estimation performance and spectral efficiency. Moreover, the two-stage cascaded network architecture proposed in this paper is also more robust to channel estimation errors.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 3","pages":"1245-1258"},"PeriodicalIF":5.3,"publicationDate":"2024-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142123063","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 : 2024-01-24DOI: 10.1109/TGCN.2024.3358230
Hrishikesh Dutta;Amit Kumar Bhuyan;Subir Biswas
Efficient slot allocation and transmit-sleep scheduling is an effective access control mechanism for improving communication performance and network lifetime in resource-constrained wireless networks. In this paper, a decentralized and multi-tier framework is presented for joint slot allocation and transmit-sleep scheduling in wireless network nodes with thin energy budget. The key learning objectives of this architecture are: collision-free transmission scheduling, reducing energy consumption, and improving network performance. This is achieved using a cooperative and decentralized learning behavior of multiple Reinforcement Learning (RL) agents. The resulting architecture provides throughput-sustainable support for data flows while minimizing energy expenditure and sleep-induced packet losses. To achieve this, a concept of Context is introduced to the RL framework in order to capture network traffic dynamics. The resulting Contextual Deep Q-Learning (CDQL) model makes the system adaptive to dynamic and heterogeneous network load. It also improves energy efficiency when compared with the traditional tabular Q-learning-based approaches. The results demonstrate how this framework can be used for prioritizing application-specific requirements, namely, energy saving and communication reliability. The trade-offs among packet drop, energy expenditure, and learning convergence are studied, and an application-specific solution is proposed for managing them. The performance is compared against an existing state-of-the-art scheduling approach. Moreover, an analytical model of the system dynamics is developed and validated using simulation for arbitrary mesh topologies and traffic patterns.
{"title":"Contextual Deep Reinforcement Learning for Flow and Energy Management in Wireless Sensor and IoT Networks","authors":"Hrishikesh Dutta;Amit Kumar Bhuyan;Subir Biswas","doi":"10.1109/TGCN.2024.3358230","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3358230","url":null,"abstract":"Efficient slot allocation and transmit-sleep scheduling is an effective access control mechanism for improving communication performance and network lifetime in resource-constrained wireless networks. In this paper, a decentralized and multi-tier framework is presented for joint slot allocation and transmit-sleep scheduling in wireless network nodes with thin energy budget. The key learning objectives of this architecture are: collision-free transmission scheduling, reducing energy consumption, and improving network performance. This is achieved using a cooperative and decentralized learning behavior of multiple Reinforcement Learning (RL) agents. The resulting architecture provides throughput-sustainable support for data flows while minimizing energy expenditure and sleep-induced packet losses. To achieve this, a concept of Context is introduced to the RL framework in order to capture network traffic dynamics. The resulting Contextual Deep Q-Learning (CDQL) model makes the system adaptive to dynamic and heterogeneous network load. It also improves energy efficiency when compared with the traditional tabular Q-learning-based approaches. The results demonstrate how this framework can be used for prioritizing application-specific requirements, namely, energy saving and communication reliability. The trade-offs among packet drop, energy expenditure, and learning convergence are studied, and an application-specific solution is proposed for managing them. The performance is compared against an existing state-of-the-art scheduling approach. Moreover, an analytical model of the system dynamics is developed and validated using simulation for arbitrary mesh topologies and traffic patterns.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 3","pages":"1233-1244"},"PeriodicalIF":5.3,"publicationDate":"2024-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142121601","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 : 2024-01-19DOI: 10.1109/TGCN.2024.3356065
Deepika Saxena;Ashutosh Kumar Singh
This paper addresses the pivotal challenge of achieving seamless performance in Cloud Data Centres $( mathbb {CDC}text{s}$ ) while meeting high availability, security, and sustainability requirements. Existing approaches often struggle to cater to all critical objectives simultaneously and overlook the significance of inter-dependent Virtual Machines (VMs) during resource distribution. To tackle these issues, a novel sustainable resource management model is proposed to provide high availability and reduce security breaches within $ mathbb {CDC}text{s}$ . The contributions include computing VM ranks to prioritize critical VMs for high availability, workload distribution with power and heat constraints for a sustainable environment, and minimizing security breaches through monitoring and terminating malicious VMs. Real-world Google Cluster workloads validate the model’s efficacy, showcasing improved availability, resource utilization, Power Usage Effectiveness (PUE), up to 15.11%, 19%, and 23.4%, respectively with reduced security breaches, and energy consumption up to 53.8% and 17.1%, respectively.
{"title":"A High Up-Time and Security Centered Resource Provisioning Model Toward Sustainable Cloud Service Management","authors":"Deepika Saxena;Ashutosh Kumar Singh","doi":"10.1109/TGCN.2024.3356065","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3356065","url":null,"abstract":"This paper addresses the pivotal challenge of achieving seamless performance in Cloud Data Centres \u0000<inline-formula> <tex-math>$( mathbb {CDC}text{s}$ </tex-math></inline-formula>\u0000) while meeting high availability, security, and sustainability requirements. Existing approaches often struggle to cater to all critical objectives simultaneously and overlook the significance of inter-dependent Virtual Machines (VMs) during resource distribution. To tackle these issues, a novel sustainable resource management model is proposed to provide high availability and reduce security breaches within \u0000<inline-formula> <tex-math>$ mathbb {CDC}text{s}$ </tex-math></inline-formula>\u0000. The contributions include computing VM ranks to prioritize critical VMs for high availability, workload distribution with power and heat constraints for a sustainable environment, and minimizing security breaches through monitoring and terminating malicious VMs. Real-world Google Cluster workloads validate the model’s efficacy, showcasing improved availability, resource utilization, Power Usage Effectiveness (PUE), up to 15.11%, 19%, and 23.4%, respectively with reduced security breaches, and energy consumption up to 53.8% and 17.1%, respectively.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 3","pages":"1182-1195"},"PeriodicalIF":5.3,"publicationDate":"2024-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142123062","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 : 2024-01-18DOI: 10.1109/TGCN.2024.3355894
Thanh V. Pham;Anh T. Pham;Susumu Ishihara
This paper studies the design of energy-efficient artificial noise (AN) schemes in the context of physical layer security in visible light communications (VLC). Two different transmission schemes termed selective AN-aided single-input single-output (SISO) and AN-aided multiple-input single-output (MISO) are examined and compared in terms of secrecy energy efficiency (SEE). In the former, the closest LED luminaire to the legitimate user (Bob) is the information-bearing signal’s transmitter. At the same time, the rest of the luminaries act as jammers transmitting AN to degrade the channels of eavesdroppers (Eves). In the latter, the information-bearing signal and AN are combined and transmitted by all luminaries. When Eves’ CSI is unknown, an indirect design to improve the SEE is formulated by maximizing Bob’s channel’s energy efficiency. A low-complexity design based on the zero-forcing criterion is also proposed. In the case of known Eves’ CSI, we study the design that maximizes the minimum SEE among those corresponding to all eavesdroppers. At their respective optimal SEEs, simulation results reveal that when Eves’ CSI is unknown, the selective AN-aided SISO transmission can archive twice as good SEE as the AN-aided MISO does. In contrast, when Eves’ CSI is known, the AN-aided MISO outperforms by 30%.
本文在可见光通信（VLC）物理层安全的背景下，研究了高能效人工噪声（AN）方案的设计。本文研究了两种不同的传输方案，分别称为选择性人工噪音辅助单输入单输出（SISO）和人工噪音辅助多输入单输出（MISO），并从保密能效（SEE）的角度进行了比较。在前者中，离合法用户（Bob）最近的 LED 灯具是信息信号的发射器。同时，其他灯具充当干扰器，发射 AN 以削弱窃听者（Eves）的信道。在后一种情况下，信息信号和 AN 合并后由所有灯具发射。当 Eves 的 CSI 未知时，可以通过最大化 Bob 信道的能效来间接设计改进 SEE。此外，还提出了一种基于零强迫准则的低复杂度设计。在已知 Eves CSI 的情况下，我们研究了在所有窃听者对应的设计中使最小 SEE 最大化的设计。仿真结果表明，在各自的最佳 SEE 下，当 Eves 的 CSI 未知时，选择性 AN 辅助 SISO 传输的 SEE 是 AN 辅助 MISO 的两倍。相比之下，当已知 Eves 的 CSI 时，AN 辅助 MISO 的性能要高出 30%。
{"title":"Design of Energy-Efficient Artificial Noise for Physical Layer Security in Visible Light Communications","authors":"Thanh V. Pham;Anh T. Pham;Susumu Ishihara","doi":"10.1109/TGCN.2024.3355894","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3355894","url":null,"abstract":"This paper studies the design of energy-efficient artificial noise (AN) schemes in the context of physical layer security in visible light communications (VLC). Two different transmission schemes termed selective AN-aided single-input single-output (SISO) and AN-aided multiple-input single-output (MISO) are examined and compared in terms of secrecy energy efficiency (SEE). In the former, the closest LED luminaire to the legitimate user (Bob) is the information-bearing signal’s transmitter. At the same time, the rest of the luminaries act as jammers transmitting AN to degrade the channels of eavesdroppers (Eves). In the latter, the information-bearing signal and AN are combined and transmitted by all luminaries. When Eves’ CSI is unknown, an indirect design to improve the SEE is formulated by maximizing Bob’s channel’s energy efficiency. A low-complexity design based on the zero-forcing criterion is also proposed. In the case of known Eves’ CSI, we study the design that maximizes the minimum SEE among those corresponding to all eavesdroppers. At their respective optimal SEEs, simulation results reveal that when Eves’ CSI is unknown, the selective AN-aided SISO transmission can archive twice as good SEE as the AN-aided MISO does. In contrast, when Eves’ CSI is known, the AN-aided MISO outperforms by 30%.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"741-755"},"PeriodicalIF":4.8,"publicationDate":"2024-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078734","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}
Backscatter communication, as an important technique in green Internet of Things, has been concerned by academic and industry to improve system capacity and simultaneously reduce network cost in a low-power-consumption way. In this paper, a sum-throughput maximization resource allocation (RA) problem is studied for a full-duplex-enhanced wireless-powered backscatter communication network, where one hybrid access point (HAP) with constant power supply can coordinate wireless energy and information transmission for multiple backscatter users without other energy sources. All users first harvest the wireless energy from the HAP during the downlink transmission and simultaneously backscatter their information to the HAP, and then send their information to the HAP during uplink transmission. Then, a sum-throughput maximization RA problem is formulated by jointly optimizing the beamforming vector of the HAP, energy-harvesting (EH) time, reflection coefficients, and the transmit power of users, where the constraints of the maximum transmit power imposed by the HAP, the minimum throughput and the EH requirement of each user are considered simultaneously. Finally, the non-convex problem is converted into a convex one by applying a series of convex optimization methods, then an iterative-based RA algorithm is proposed to solve it. Simulation results verify the effectiveness of the proposed algorithm.
后向散射通信作为绿色物联网的一项重要技术，一直受到学术界和产业界的关注，它能以低功耗的方式提高系统容量，同时降低网络成本。本文研究了一个全双工增强型无线供电反向散射通信网络的总吞吐量最大化资源分配（RA）问题，在该网络中，一个恒定供电的混合接入点（HAP）可以在没有其他能源的情况下协调多个反向散射用户的无线能量和信息传输。所有用户首先在下行链路传输过程中从混合接入点获取无线能量，同时向混合接入点反向散射信息，然后在上行链路传输过程中向混合接入点发送信息。然后，通过联合优化 HAP 的波束成形向量、能量收集（EH）时间、反射系数和用户的发射功率，提出了总吞吐量最大化 RA 问题，其中同时考虑了 HAP 的最大发射功率、最小吞吐量和每个用户的 EH 要求等约束条件。最后，通过应用一系列凸优化方法将非凸问题转化为凸问题，并提出了一种基于迭代的 RA 算法来解决该问题。仿真结果验证了所提算法的有效性。
{"title":"Full-Duplex-Enhanced Wireless-Powered Backscatter Communication Networks: Radio Resource Allocation and Beamforming Joint Optimization","authors":"Xiaoxi Zhang;Yongjun Xu;Haibo Zhang;Gongpu Wang;Xingwang Li;Chau Yuen","doi":"10.1109/TGCN.2024.3354986","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3354986","url":null,"abstract":"Backscatter communication, as an important technique in green Internet of Things, has been concerned by academic and industry to improve system capacity and simultaneously reduce network cost in a low-power-consumption way. In this paper, a sum-throughput maximization resource allocation (RA) problem is studied for a full-duplex-enhanced wireless-powered backscatter communication network, where one hybrid access point (HAP) with constant power supply can coordinate wireless energy and information transmission for multiple backscatter users without other energy sources. All users first harvest the wireless energy from the HAP during the downlink transmission and simultaneously backscatter their information to the HAP, and then send their information to the HAP during uplink transmission. Then, a sum-throughput maximization RA problem is formulated by jointly optimizing the beamforming vector of the HAP, energy-harvesting (EH) time, reflection coefficients, and the transmit power of users, where the constraints of the maximum transmit power imposed by the HAP, the minimum throughput and the EH requirement of each user are considered simultaneously. Finally, the non-convex problem is converted into a convex one by applying a series of convex optimization methods, then an iterative-based RA algorithm is proposed to solve it. Simulation results verify the effectiveness of the proposed algorithm.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"730-740"},"PeriodicalIF":4.8,"publicationDate":"2024-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078767","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 : 2024-01-10DOI: 10.1109/TGCN.2024.3352173
Jiansong Miao;Shanling Bai;Shahid Mumtaz;Qian Zhang;Junsheng Mu
The integration of unmanned aerial vehicles (UAVs) in future communication networks has received great attention, and it plays an essential role in many applications, such as military reconnaissance, fire monitoring, etc. In this paper, we consider a UAV-aided video transmission system based on mobile edge computing (MEC). Considering the short latency requirements, the UAV acts as a MEC server to transcode the videos and as a relay to forward the transcoded videos to the ground base station. Subject to constraints on discrete variables and short latency, we aim to maximize the cumulative utility by jointly optimizing the power allocation, video transcoding policy, computational resources allocation, and UAV flight trajectory. The above non-convex optimization problem is modeled as a Markov decision process (MDP) and solved by a deep deterministic policy gradient (DDPG) algorithm to realize continuous action control by policy iteration. Simulation results show that the DDPG algorithm performs better than deep Q-learning network algorithm (DQN) and actor-critic (AC) algorithm.
{"title":"Utility-Oriented Optimization for Video Streaming in UAV-Aided MEC Network: A DRL Approach","authors":"Jiansong Miao;Shanling Bai;Shahid Mumtaz;Qian Zhang;Junsheng Mu","doi":"10.1109/TGCN.2024.3352173","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3352173","url":null,"abstract":"The integration of unmanned aerial vehicles (UAVs) in future communication networks has received great attention, and it plays an essential role in many applications, such as military reconnaissance, fire monitoring, etc. In this paper, we consider a UAV-aided video transmission system based on mobile edge computing (MEC). Considering the short latency requirements, the UAV acts as a MEC server to transcode the videos and as a relay to forward the transcoded videos to the ground base station. Subject to constraints on discrete variables and short latency, we aim to maximize the cumulative utility by jointly optimizing the power allocation, video transcoding policy, computational resources allocation, and UAV flight trajectory. The above non-convex optimization problem is modeled as a Markov decision process (MDP) and solved by a deep deterministic policy gradient (DDPG) algorithm to realize continuous action control by policy iteration. Simulation results show that the DDPG algorithm performs better than deep Q-learning network algorithm (DQN) and actor-critic (AC) algorithm.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"878-889"},"PeriodicalIF":4.8,"publicationDate":"2024-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078817","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 : 2024-01-08DOI: 10.1109/TGCN.2024.3350787
Hao Lin;Mustafa A. Kishk;Mohamed-Slim Alouini
Bridging the digital divide is one of the goals of mobile networks in the future, and further building IoT networks in rural areas is a feasible solution. This paper studies the downlink performance of rural wireless networks, where IoT devices we consider are battery-less and powered only by ambient radio-frequency (RF) signals. We model a rural area as a finite area that is far from the city center. The base stations (BSs) in the whole city and the access points (APs) in the finite network both act as sources of wireless RF signals harvested by IoT devices. We assume that BSs follow an inhomogeneous Poisson Point Process (PPP) with a 2D-Gaussian density, and a fixed number of APs are uniformly distributed inside the finite area following a Binomial Point Process (BPP). The IoT devices we consider can harvest energy and receive downlink signals in each time slot, which is divided into two parts: (1) a charging sub-slot, where the RF signals from BSs and APs are harvested by IoT devices, and (2) a transmission sub-slot, where each IoT device uses the harvested energy to receive and process downlink signals. We consider two main system requirements: minimum energy requirement and signal-to-interference-plus-noise ratio (SINR). Using these two parameters, we investigate the overall coverage probability (OCP) related to them. We first study the effect of remoteness in rural areas on energy harvesting performance. Then we analyze the influence of IoT device’s location and the number of APs on coverage probability when the effect of BSs can be ignored. This paper shows that the IoT devices located inside the rural area can obtain about twice the ECP and OCP of IoT devices located near the edge. For the average downlink performance in rural areas with radii less than 100 m, more than 80% of the RF-powered IoT devices can be supported when there are 100 APs deployed.
{"title":"Performance Evaluation of RF-Powered IoT in Rural Areas: The Wireless Power Digital Divide","authors":"Hao Lin;Mustafa A. Kishk;Mohamed-Slim Alouini","doi":"10.1109/TGCN.2024.3350787","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3350787","url":null,"abstract":"Bridging the digital divide is one of the goals of mobile networks in the future, and further building IoT networks in rural areas is a feasible solution. This paper studies the downlink performance of rural wireless networks, where IoT devices we consider are battery-less and powered only by ambient radio-frequency (RF) signals. We model a rural area as a finite area that is far from the city center. The base stations (BSs) in the whole city and the access points (APs) in the finite network both act as sources of wireless RF signals harvested by IoT devices. We assume that BSs follow an inhomogeneous Poisson Point Process (PPP) with a 2D-Gaussian density, and a fixed number of APs are uniformly distributed inside the finite area following a Binomial Point Process (BPP). The IoT devices we consider can harvest energy and receive downlink signals in each time slot, which is divided into two parts: (1) a charging sub-slot, where the RF signals from BSs and APs are harvested by IoT devices, and (2) a transmission sub-slot, where each IoT device uses the harvested energy to receive and process downlink signals. We consider two main system requirements: minimum energy requirement and signal-to-interference-plus-noise ratio (SINR). Using these two parameters, we investigate the overall coverage probability (OCP) related to them. We first study the effect of remoteness in rural areas on energy harvesting performance. Then we analyze the influence of IoT device’s location and the number of APs on coverage probability when the effect of BSs can be ignored. This paper shows that the IoT devices located inside the rural area can obtain about twice the ECP and OCP of IoT devices located near the edge. For the average downlink performance in rural areas with radii less than 100 m, more than 80% of the RF-powered IoT devices can be supported when there are 100 APs deployed.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"716-729"},"PeriodicalIF":4.8,"publicationDate":"2024-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078733","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 : 2024-01-08DOI: 10.1109/TGCN.2024.3350735
Abdullatif Albaseer;Abegaz Mohammed Seid;Mohamed Abdallah;Ala Al-Fuqaha;Aiman Erbad
Researchers and practitioners have recently shown interest in deploying federated learning for enhanced privacy preservation in wireless edge networks. In such settings, resource-constrained user equipment (UE) often experiences unfair energy consumption and performance degradation of machine learning models due to data heterogeneity and constrained computation and communication resources. Several approaches have been proposed in the literature to reduce energy consumption, including scheduling a subset of UEs to undertake learning tasks based on their energy budgets. However, these approaches are inherently unfair as the frequently selected UEs rapidly deplete their energy and are rendered inaccessible. Furthermore, the server may be unable to capture the incongruent data distribution, resulting in a biased model. In this paper, we propose a novel approach that addresses those challenges, considering the historical participation of the UEs to ensure that all the training data of the UEs are incorporated into the global model. Specifically, using Jain’s fairness index, we formulate the overall optimization problem, decompose it into two sub-problems, and iteratively solve the sub-problems. Towards this end, we partition the optimization variables into two-blocks; one on the server-side and another on the UEs’ side. The server-side algorithm aims to balance energy usage and learning performance, while the client-side algorithm seeks to optimize CPU frequency and transmit power. Extensive experiments using two realistic datasets, FEMNIST and CIFAR-10, indicate that the proposed algorithms minimize overall energy while curbing unfair energy consumption between the UEs, accelerating convergence rates, and significantly enhancing local accuracy for all UEs.
{"title":"Novel Approach for Curbing Unfair Energy Consumption and Biased Model in Federated Edge Learning","authors":"Abdullatif Albaseer;Abegaz Mohammed Seid;Mohamed Abdallah;Ala Al-Fuqaha;Aiman Erbad","doi":"10.1109/TGCN.2024.3350735","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3350735","url":null,"abstract":"Researchers and practitioners have recently shown interest in deploying federated learning for enhanced privacy preservation in wireless edge networks. In such settings, resource-constrained user equipment (UE) often experiences unfair energy consumption and performance degradation of machine learning models due to data heterogeneity and constrained computation and communication resources. Several approaches have been proposed in the literature to reduce energy consumption, including scheduling a subset of UEs to undertake learning tasks based on their energy budgets. However, these approaches are inherently unfair as the frequently selected UEs rapidly deplete their energy and are rendered inaccessible. Furthermore, the server may be unable to capture the incongruent data distribution, resulting in a biased model. In this paper, we propose a novel approach that addresses those challenges, considering the historical participation of the UEs to ensure that all the training data of the UEs are incorporated into the global model. Specifically, using Jain’s fairness index, we formulate the overall optimization problem, decompose it into two sub-problems, and iteratively solve the sub-problems. Towards this end, we partition the optimization variables into two-blocks; one on the server-side and another on the UEs’ side. The server-side algorithm aims to balance energy usage and learning performance, while the client-side algorithm seeks to optimize CPU frequency and transmit power. Extensive experiments using two realistic datasets, FEMNIST and CIFAR-10, indicate that the proposed algorithms minimize overall energy while curbing unfair energy consumption between the UEs, accelerating convergence rates, and significantly enhancing local accuracy for all UEs.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"865-877"},"PeriodicalIF":4.8,"publicationDate":"2024-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078855","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 : 2024-01-04DOI: 10.1109/TGCN.2024.3349697
Alexander Herzog;Robbie Southam;Othmane Belarbi;Saif Anwar;Marcello Bullo;Pietro Carnelli;Aftab Khan
Federated Learning (FL) is fast becoming one of the most prevalent distributed learning techniques focused on privacy preservation and communication efficiency for large-scale Internet of Things (IoT) deployments. FL is a distributed learning approach to training models on distributed devices. Since local data remains on-device, communication through the network is reduced. However, in large-scale IoT environments or resource constrained networks, typical FL approaches significantly suffer in performance due to longer communication times. In this paper, we propose two methods for further reducing communication volume in resource restricted FL deployments. In our first method, which we term Selective Updates (SU), local models are trained until a dynamic threshold on model performance is surpassed before sending updates to a centralised Parameter Server (PS). This allows for minimal updates being transmitted, thus reducing communication overheads. Our second method, Adaptive Masking (AM), performs parameter masking on both the global and local models prior to sharing. With AM, we select model parameters that have changed the most between training rounds. We extensively evaluate our proposed methods against state-of-the-art communication reduction strategies using two common benchmark datasets, and under different communication constrained settings. Our proposed methods reduce the overall communication volume by over 20%, without affecting the model accuracy.
{"title":"Selective Updates and Adaptive Masking for Communication-Efficient Federated Learning","authors":"Alexander Herzog;Robbie Southam;Othmane Belarbi;Saif Anwar;Marcello Bullo;Pietro Carnelli;Aftab Khan","doi":"10.1109/TGCN.2024.3349697","DOIUrl":"https://doi.org/10.1109/TGCN.2024.3349697","url":null,"abstract":"Federated Learning (FL) is fast becoming one of the most prevalent distributed learning techniques focused on privacy preservation and communication efficiency for large-scale Internet of Things (IoT) deployments. FL is a distributed learning approach to training models on distributed devices. Since local data remains on-device, communication through the network is reduced. However, in large-scale IoT environments or resource constrained networks, typical FL approaches significantly suffer in performance due to longer communication times. In this paper, we propose two methods for further reducing communication volume in resource restricted FL deployments. In our first method, which we term Selective Updates (SU), local models are trained until a dynamic threshold on model performance is surpassed before sending updates to a centralised Parameter Server (PS). This allows for minimal updates being transmitted, thus reducing communication overheads. Our second method, Adaptive Masking (AM), performs parameter masking on both the global and local models prior to sharing. With AM, we select model parameters that have changed the most between training rounds. We extensively evaluate our proposed methods against state-of-the-art communication reduction strategies using two common benchmark datasets, and under different communication constrained settings. Our proposed methods reduce the overall communication volume by over 20%, without affecting the model accuracy.","PeriodicalId":13052,"journal":{"name":"IEEE Transactions on Green Communications and Networking","volume":"8 2","pages":"852-864"},"PeriodicalIF":4.8,"publicationDate":"2024-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141078775","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：