Pub Date : 2024-11-25DOI: 10.1109/TMC.2024.3505781
Maria Tsampazi;Salvatore D'Oro;Michele Polese;Leonardo Bonati;Gwenael Poitau;Michael Healy;Mohammad Alavirad;Tommaso Melodia
The highly heterogeneous ecosystem of Next Generation (NextG) wireless communication systems calls for novel networking paradigms where functionalities and operations can be dynamically and optimally reconfigured in real time to adapt to changing traffic conditions and satisfy stringent and diverse Quality of Service (QoS) demands. Open Radio Access Network (RAN) technologies, and specifically those being standardized by the O-RAN Alliance, make it possible to integrate network intelligence into the once monolithic RAN via intelligent applications, namely, xApps and rApps. These applications enable flexible control of the network resources and functionalities, network management, and orchestration through data-driven intelligent control loops. Recent work has showed how Deep Reinforcement Learning (DRL) is effective in dynamically controlling O-RAN systems. However, how to design these solutions in a way that manages heterogeneous optimization goals and prevents unfair resource allocation is still an open challenge, with the logic within DRL agents often considered as a opaque system. In this paper, we introduce PandORA, a framework to automatically design and train DRL agents for Open RAN applications, package them as xApps and evaluate them in the Colosseum wireless network emulator. We benchmark 23 xApps that embed DRL agents trained using different architectures, reward design, action spaces, and decision-making timescales, and with the ability to hierarchically control different network parameters. We test these agents on the Colosseum testbed under diverse traffic and channel conditions, in static and mobile setups. Our experimental results indicate how suitable fine-tuning of the RAN control timers, as well as proper selection of reward designs and DRL architectures can boost network performance according to the network conditions and demand. Notably, finer decision-making granularities can improve Massive Machine-Type Communications (mMTC)’s performance by $sim! 56%$ and even increase Enhanced Mobile Broadband (eMBB) Throughput by $sim! 99%$.
{"title":"PandORA: Automated Design and Comprehensive Evaluation of Deep Reinforcement Learning Agents for Open RAN","authors":"Maria Tsampazi;Salvatore D'Oro;Michele Polese;Leonardo Bonati;Gwenael Poitau;Michael Healy;Mohammad Alavirad;Tommaso Melodia","doi":"10.1109/TMC.2024.3505781","DOIUrl":"https://doi.org/10.1109/TMC.2024.3505781","url":null,"abstract":"The highly heterogeneous ecosystem of Next Generation (NextG) wireless communication systems calls for novel networking paradigms where functionalities and operations can be dynamically and optimally reconfigured in real time to adapt to changing traffic conditions and satisfy stringent and diverse Quality of Service (QoS) demands. Open Radio Access Network (RAN) technologies, and specifically those being standardized by the O-RAN Alliance, make it possible to integrate network intelligence into the once monolithic RAN via intelligent applications, namely, xApps and rApps. These applications enable flexible control of the network resources and functionalities, network management, and orchestration through data-driven intelligent control loops. Recent work has showed how Deep Reinforcement Learning (DRL) is effective in dynamically controlling O-RAN systems. However, how to design these solutions in a way that manages heterogeneous optimization goals and prevents unfair resource allocation is still an open challenge, with the logic within DRL agents often considered as a opaque system. In this paper, we introduce PandORA, a framework to automatically design and train DRL agents for Open RAN applications, package them as xApps and evaluate them in the Colosseum wireless network emulator. We benchmark 23 xApps that embed DRL agents trained using different architectures, reward design, action spaces, and decision-making timescales, and with the ability to hierarchically control different network parameters. We test these agents on the Colosseum testbed under diverse traffic and channel conditions, in static and mobile setups. Our experimental results indicate how suitable fine-tuning of the RAN control timers, as well as proper selection of reward designs and DRL architectures can boost network performance according to the network conditions and demand. Notably, finer decision-making granularities can improve Massive Machine-Type Communications (mMTC)’s performance by <inline-formula><tex-math>$sim! 56%$</tex-math></inline-formula> and even increase Enhanced Mobile Broadband (eMBB) Throughput by <inline-formula><tex-math>$sim! 99%$</tex-math></inline-formula>.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"3223-3240"},"PeriodicalIF":7.7,"publicationDate":"2024-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143564146","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 : 2024-11-22DOI: 10.1109/TMC.2024.3505155
Zemin Sun;Geng Sun;Qingqing Wu;Long He;Shuang Liang;Hongyang Pan;Dusit Niyato;Chau Yuen;Victor C. M. Leung
Unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) is emerging as a promising paradigm to provide aerial-terrestrial computing services in close proximity to mobile devices (MDs). However, meeting the demands of computation-intensive and delay-sensitive tasks for MDs poses several challenges, including the demand-supply contradiction between MDs and MEC servers, the demand-supply discrepancy between MDs and MEC servers, the trajectory control requirements on energy efficiency and timeliness, and the different time-scale dynamics of the network. To address these issues, we first present a hierarchical architecture by incorporating terrestrial-aerial computing capabilities and leveraging UAV flexibility. Furthermore, we formulate a joint computing resource allocation, computation offloading, and trajectory control problem to maximize the system utility. Since the problem is a non-convex and NP-hard mixed integer nonlinear programming (MINLP), we propose a two-timescale joint computing resource allocation, computation offloading, and trajectory control (TJCCT) approach for solving the problem. In the short timescale, we propose a price-incentive model for on-demand computing resource allocation and a matching mechanism-based method for computation offloading. In the long timescale, we propose a convex optimization-based method for UAV trajectory control. Besides, we theoretically prove the stability and polynomial complexity of TJCCT. Extensive simulation results demonstrate that the proposed TJCCT is able to achieve superior performances in terms of the system utility, average processing rate, average completion delay, average completion ratio, and average cost, while meeting the energy constraints despite the trade-off of the increased energy consumption.
{"title":"TJCCT: A Two-Timescale Approach for UAV-Assisted Mobile Edge Computing","authors":"Zemin Sun;Geng Sun;Qingqing Wu;Long He;Shuang Liang;Hongyang Pan;Dusit Niyato;Chau Yuen;Victor C. M. Leung","doi":"10.1109/TMC.2024.3505155","DOIUrl":"https://doi.org/10.1109/TMC.2024.3505155","url":null,"abstract":"Unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) is emerging as a promising paradigm to provide aerial-terrestrial computing services in close proximity to mobile devices (MDs). However, meeting the demands of computation-intensive and delay-sensitive tasks for MDs poses several challenges, including the demand-supply contradiction between MDs and MEC servers, the demand-supply discrepancy between MDs and MEC servers, the trajectory control requirements on energy efficiency and timeliness, and the different time-scale dynamics of the network. To address these issues, we first present a hierarchical architecture by incorporating terrestrial-aerial computing capabilities and leveraging UAV flexibility. Furthermore, we formulate a joint computing resource allocation, computation offloading, and trajectory control problem to maximize the system utility. Since the problem is a non-convex and NP-hard mixed integer nonlinear programming (MINLP), we propose a two-timescale joint computing resource allocation, computation offloading, and trajectory control (TJCCT) approach for solving the problem. In the short timescale, we propose a price-incentive model for on-demand computing resource allocation and a matching mechanism-based method for computation offloading. In the long timescale, we propose a convex optimization-based method for UAV trajectory control. Besides, we theoretically prove the stability and polynomial complexity of TJCCT. Extensive simulation results demonstrate that the proposed TJCCT is able to achieve superior performances in terms of the system utility, average processing rate, average completion delay, average completion ratio, and average cost, while meeting the energy constraints despite the trade-off of the increased energy consumption.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"3130-3147"},"PeriodicalIF":7.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563921","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 : 2024-11-22DOI: 10.1109/TMC.2024.3504960
Ming Xia;Min Huang;Qiuqi Pan;Yunhan Wang;Xiaoyan Wang;Kaikai Chi
This article presents Insight, an AR system for visualizing the IoT-enabled smart environment without relying on the unique appearances, barcodes, world coordinates, or wireless signals of IoT infrastructures. The system analyzes the camera video and motion data taken by mobile AR equipment to extract the self and cross visual geometries describing the poses and geographic distribution of nearby IoT devices. To recognize IoT devices using the extracted geometries, Insight operates in two phases. At deployment time, it learns pairwise mappings from the visual geometries to the corresponding device identities. After that, it leverages the geometries scanned at run time to look for a partial assignment to the recorded geometries, allowing it to automatically recognize the IoT devices in AR view. As such, our system turns the IoT device recognition task into a geometry matching problem, which is further formalized as to perform Subset, Incomplete, and Duplicated Point Cloud Registration (SID-PCR) in this work. We design a deep neural network paying specific edge- and spectral-wise graph attention to solve SID-PCR, and implement a prototype that adaptively requests visual geometry scan and registration operations for accurate recognition. The performance of Insight is validated using both synthetic data and a real-world testbed.
{"title":"Visualizing the Smart Environment in AR: An Approach Based on Visual Geometry Matching","authors":"Ming Xia;Min Huang;Qiuqi Pan;Yunhan Wang;Xiaoyan Wang;Kaikai Chi","doi":"10.1109/TMC.2024.3504960","DOIUrl":"https://doi.org/10.1109/TMC.2024.3504960","url":null,"abstract":"This article presents <sc>Insight</small>, an AR system for visualizing the IoT-enabled smart environment without relying on the unique appearances, barcodes, world coordinates, or wireless signals of IoT infrastructures. The system analyzes the camera video and motion data taken by mobile AR equipment to extract the self and cross visual geometries describing the poses and geographic distribution of nearby IoT devices. To recognize IoT devices using the extracted geometries, <sc>Insight</small> operates in two phases. At deployment time, it learns pairwise mappings from the visual geometries to the corresponding device identities. After that, it leverages the geometries scanned at run time to look for a partial assignment to the recorded geometries, allowing it to automatically recognize the IoT devices in AR view. As such, our system turns the IoT device recognition task into a geometry matching problem, which is further formalized as to perform Subset, Incomplete, and Duplicated Point Cloud Registration (SID-PCR) in this work. We design a deep neural network paying specific edge- and spectral-wise graph attention to solve SID-PCR, and implement a prototype that adaptively requests visual geometry scan and registration operations for accurate recognition. The performance of <sc>Insight</small> is validated using both synthetic data and a real-world testbed.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2900-2916"},"PeriodicalIF":7.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563902","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 : 2024-11-22DOI: 10.1109/TMC.2024.3504859
Qinglong Zhang;Rui Han;Chi Harold Liu;Guoren Wang;Song Guo;Lydia Y. Chen
Foundation models (FMs) such as large language models are becoming the backbone technology for artificial intelligence systems. It is particularly challenging to deploy multiple FMs on edge devices, which not only have limited computational resources, but also encounter unseen input data from evolving domains or learning tasks. When new data arrives, existing prior art of FM mainly focuses on retraining compressed models of predetermined network architectures, limiting the feasibility of edge devices to efficiently achieve high accuracy for FMs. In this paper, we propose EdgeTA, a neuron-grained FM scaling system to maximize the overall accuracy of FMs promptly in response to their data dynamics. EdgeTA's key design features in scaling are (i) proxy mechanism, which adaptively transforms a FM into a compact architecture retaining the most important neurons to the input data, and (ii) neuron-grained scheduler, which jointly optimizes model sizes and resource allocation for all FMs on edge devices. Under tight retraining window and limited device resources, the design of EdgeTA can achieve most of the original FM's accuracy with much smaller retraining costs. We implement EdgeTA on FMs of natural language processing, computer vision and multimodal applications. Comparison results against state-of-the-art techniques show that our approach improves accuracy by 21.88% and reduces memory footprint and energy consumptions by 27.14% and 65.65%, while further achieving 15.96% overall accuracy improvement via neuron-grained scheduling.
{"title":"EdgeTA: Neuron-Grained Scaling of Foundation Models in Edge-Side Retraining","authors":"Qinglong Zhang;Rui Han;Chi Harold Liu;Guoren Wang;Song Guo;Lydia Y. Chen","doi":"10.1109/TMC.2024.3504859","DOIUrl":"https://doi.org/10.1109/TMC.2024.3504859","url":null,"abstract":"Foundation models (FMs) such as large language models are becoming the backbone technology for artificial intelligence systems. It is particularly challenging to deploy multiple FMs on edge devices, which not only have limited computational resources, but also encounter unseen input data from evolving domains or learning tasks. When new data arrives, existing prior art of FM mainly focuses on retraining compressed models of predetermined network architectures, limiting the feasibility of edge devices to efficiently achieve high accuracy for FMs. In this paper, we propose EdgeTA, a neuron-grained FM scaling system to maximize the overall accuracy of FMs promptly in response to their data dynamics. EdgeTA's key design features in scaling are (i) proxy mechanism, which adaptively transforms a FM into a compact architecture retaining the most important neurons to the input data, and (ii) neuron-grained scheduler, which jointly optimizes model sizes and resource allocation for all FMs on edge devices. Under tight retraining window and limited device resources, the design of EdgeTA can achieve most of the original FM's accuracy with much smaller retraining costs. We implement EdgeTA on FMs of natural language processing, computer vision and multimodal applications. Comparison results against state-of-the-art techniques show that our approach improves accuracy by 21.88% and reduces memory footprint and energy consumptions by 27.14% and 65.65%, while further achieving 15.96% overall accuracy improvement via neuron-grained scheduling.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2690-2707"},"PeriodicalIF":7.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563951","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}
The video analytics system utilizes deep learning models (DNN) to perform inference on the videos captured by cameras. Continuous learning algorithms are used to address the data drift problem in video analytics systems. However, uploading images from deployment environments and processing on the cloud carry the risk of privacy leakage. In this paper, we have designed a system called CL-Shield to protect user’s privacy. First, we review the causes of privacy leakage in a continuous learning system and propose the objective of full privacy protection. Second, we design an online training mechanism based on a scene library to avoid direct uploading of user’s frames to the cloud server. Lastly, we design a fast training set search algorithm based on a novel Ebv-List, which effectively improves the speed of model updates. We collect various real-world scenario data to build our scene library and validate our system on a dataset of over 10 hours. The experiments demonstrate that our privacy-aware continuous learning system achieves an F1-score of over 92% compared to the conventional systems without protecting privacy and has long-term stability in analytic F1-score.
{"title":"CL-Shield: A Continuous Learning System for Protecting User Privacy","authors":"Tianyu Li;Hanling Wang;Qing Li;Yong Jiang;Zhenhui Yuan","doi":"10.1109/TMC.2024.3504721","DOIUrl":"https://doi.org/10.1109/TMC.2024.3504721","url":null,"abstract":"The video analytics system utilizes deep learning models (DNN) to perform inference on the videos captured by cameras. Continuous learning algorithms are used to address the data drift problem in video analytics systems. However, uploading images from deployment environments and processing on the cloud carry the risk of privacy leakage. In this paper, we have designed a system called CL-Shield to protect user’s privacy. First, we review the causes of privacy leakage in a continuous learning system and propose the objective of full privacy protection. Second, we design an online training mechanism based on a scene library to avoid direct uploading of user’s frames to the cloud server. Lastly, we design a fast training set search algorithm based on a novel Ebv-List, which effectively improves the speed of model updates. We collect various real-world scenario data to build our scene library and validate our system on a dataset of over 10 hours. The experiments demonstrate that our privacy-aware continuous learning system achieves an F1-score of over 92% compared to the conventional systems without protecting privacy and has long-term stability in analytic F1-score.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"3148-3162"},"PeriodicalIF":7.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563897","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}
With the proliferation of mobile terminals, the continuous upgrading of services, 4G LTE networks are showing signs of weakness. To enhance the capacity of wireless networks, millimeter waves are introduced to drive the evolution of networks towards multi-band 5G heterogeneous networks. The distinct propagation characteristics of mmWaves, microwaves, as well as the vastly different hardware configurations of heterogeneous base stations, make traditional access strategies no longer effective. Therefore, to narrowing the gap between theory, practice, we investigate the access strategy in multi-band 5G heterogeneous networks, taking into account the characteristics of mobile users, asynchronous switching between uplink, downlink of pico base stations, asymmetric service requirements, user communication continuity. We formulate the problem as integer nonlinear programming, prove its intractability. Thereby, we decouple it into three subproblems: user association, switch point selection, subchannel allocation, design an algorithm based on optimal matching, spectral clustering to solve it efficiently. The simulation results show that the proposed algorithm outperforms the comparison methods in terms of overall data rate, effective data rate, number of satisfied users.
{"title":"User Association and Channel Allocation in 5G Mobile Asymmetric Multi-Band Heterogeneous Networks","authors":"Miao Dai;Gang Sun;Hongfang Yu;Sheng Wang;Dusit Niyato","doi":"10.1109/TMC.2024.3503632","DOIUrl":"https://doi.org/10.1109/TMC.2024.3503632","url":null,"abstract":"With the proliferation of mobile terminals, the continuous upgrading of services, 4G LTE networks are showing signs of weakness. To enhance the capacity of wireless networks, millimeter waves are introduced to drive the evolution of networks towards multi-band 5G heterogeneous networks. The distinct propagation characteristics of mmWaves, microwaves, as well as the vastly different hardware configurations of heterogeneous base stations, make traditional access strategies no longer effective. Therefore, to narrowing the gap between theory, practice, we investigate the access strategy in multi-band 5G heterogeneous networks, taking into account the characteristics of mobile users, asynchronous switching between uplink, downlink of pico base stations, asymmetric service requirements, user communication continuity. We formulate the problem as integer nonlinear programming, prove its intractability. Thereby, we decouple it into three subproblems: user association, switch point selection, subchannel allocation, design an algorithm based on optimal matching, spectral clustering to solve it efficiently. The simulation results show that the proposed algorithm outperforms the comparison methods in terms of overall data rate, effective data rate, number of satisfied users.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"3092-3109"},"PeriodicalIF":7.7,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563899","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 : 2024-11-21DOI: 10.1109/TMC.2024.3503682
Siyang Guo;Yaming Guo;Hui Zhang;Junbo Wang
The increasingly popular federated learning still faces the practical challenge of non-independent and identically distributed data. Most efforts to address this issue focus on limiting local updates or enhancing model aggregation. However, these methods either restrict the learning capacity of local models or overlook the negative knowledge transfer caused by local objective divergences. In contrast, we observe that the global update can be re-expressed as a weighted sum of the gradients of samples from different classes. Therefore, we hypothesize that the competition among local updates may arise from the conflict between the gradients of samples belonging to different classes. Inspired by this insight, we introduce the novel perspective of orthogonal class gradients, aimed at eliminating interference between updates from different classes without the aforementioned drawbacks. To this end, this paper presents FedOCF, which implements orthogonal class gradient constraints by encouraging orthogonality among features of different classes. Specifically, FedOCF maintains a generator to learn features that are orthogonal for different classes and utilizes it to regularize features learned in local learning. Theoretically, we also demonstrate that FedOCF can improve generalization performance through feature conditional distribution alignment during local learning. Extensive experiments validate the excellent performance of FedOCF in various heterogeneous scenarios.
{"title":"Mitigating Update Conflict in Non-IID Federated Learning via Orthogonal Class Gradients","authors":"Siyang Guo;Yaming Guo;Hui Zhang;Junbo Wang","doi":"10.1109/TMC.2024.3503682","DOIUrl":"https://doi.org/10.1109/TMC.2024.3503682","url":null,"abstract":"The increasingly popular federated learning still faces the practical challenge of non-independent and identically distributed data. Most efforts to address this issue focus on limiting local updates or enhancing model aggregation. However, these methods either restrict the learning capacity of local models or overlook the negative knowledge transfer caused by local objective divergences. In contrast, we observe that the global update can be re-expressed as a weighted sum of the gradients of samples from different classes. Therefore, we hypothesize that the competition among local updates may arise from the conflict between the gradients of samples belonging to different classes. Inspired by this insight, we introduce the novel perspective of orthogonal class gradients, aimed at eliminating interference between updates from different classes without the aforementioned drawbacks. To this end, this paper presents <sc>FedOCF</small>, which implements orthogonal class gradient constraints by encouraging orthogonality among features of different classes. Specifically, <sc>FedOCF</small> maintains a generator to learn features that are orthogonal for different classes and utilizes it to regularize features learned in local learning. Theoretically, we also demonstrate that <sc>FedOCF</small> can improve generalization performance through feature conditional distribution alignment during local learning. Extensive experiments validate the excellent performance of <sc>FedOCF</small> in various heterogeneous scenarios.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2967-2978"},"PeriodicalIF":7.7,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563967","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 : 2024-11-21DOI: 10.1109/TMC.2024.3504284
Rongwei Lu;Yutong Jiang;Yinan Mao;Chen Tang;Bin Chen;Laizhong Cui;Zhi Wang
Federated Learning (FL) in mobile environments faces significant communication bottlenecks. Gradient compression has proven as an effective solution to this issue, offering substantial benefits in environments with limited bandwidth and metered data. Yet, it encounters severe performance drops in non-IID environments due to a one-size-fits-all compression approach, which does not account for the varying data volumes across workers. Assigning varying compression ratios to workers with distinct data distributions and volumes is therefore a promising solution. This work derives the convergence rate of distributed SGD with non-uniform compression, which reveals the intricate relationship between model convergence and the compression ratios applied to individual workers. Accordingly, we frame the relative compression ratio assignment as an $n$-variable chi-squared nonlinear optimization problem, constrained by a limited communication budget. We propose DAGC-R, which assigns conservative compression to workers handling larger data volumes. Recognizing the computational limitations of mobile devices, we propose the DAGC-A, which is computationally less demanding and enhances the robustness of compression in non-IID scenarios. Our experiments confirm that the DAGC-R and DAGC-A can speed up the training speed by up to 25.43% and 16.65% compared to the uniform compression respectively, when dealing with highly imbalanced data volume distribution and restricted communication.
{"title":"Data-Aware Gradient Compression for FL in Communication-Constrained Mobile Computing","authors":"Rongwei Lu;Yutong Jiang;Yinan Mao;Chen Tang;Bin Chen;Laizhong Cui;Zhi Wang","doi":"10.1109/TMC.2024.3504284","DOIUrl":"https://doi.org/10.1109/TMC.2024.3504284","url":null,"abstract":"Federated Learning (FL) in mobile environments faces significant communication bottlenecks. Gradient compression has proven as an effective solution to this issue, offering substantial benefits in environments with limited bandwidth and metered data. Yet, it encounters severe performance drops in non-IID environments due to a one-size-fits-all compression approach, which does not account for the varying data volumes across workers. Assigning varying compression ratios to workers with distinct data distributions and volumes is therefore a promising solution. This work derives the convergence rate of distributed SGD with non-uniform compression, which reveals the intricate relationship between model convergence and the compression ratios applied to individual workers. Accordingly, we frame the relative compression ratio assignment as an <inline-formula><tex-math>$n$</tex-math></inline-formula>-variable chi-squared nonlinear optimization problem, constrained by a limited communication budget. We propose DAGC-R, which assigns conservative compression to workers handling larger data volumes. Recognizing the computational limitations of mobile devices, we propose the DAGC-A, which is computationally less demanding and enhances the robustness of compression in non-IID scenarios. Our experiments confirm that the DAGC-R and DAGC-A can speed up the training speed by up to 25.43% and 16.65% compared to the uniform compression respectively, when dealing with highly imbalanced data volume distribution and restricted communication.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2755-2768"},"PeriodicalIF":7.7,"publicationDate":"2024-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143564177","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 : 2024-11-20DOI: 10.1109/TMC.2024.3502683
Yanan Li;Penghong Zhao;Xiao Ma;Haitao Yuan;Zhe Fu;Mengwei Xu;Shangguang Wang
With the rapid development of edge computing in the post-COVID19 pandemic period, precise workload forecasting is considered the basis for making full use of the edge-limited resources, and both edge service providers (ESPs) and edge service consumers (ESCs) can benefit significantly from it. Existing paradigms of workload forecasting (i.e., edge-only or cloud-only) are improper, due to failing to consider the inter-site correlations and might suffer from significant data transmission delays. With the increasing adoption of edge platforms by web services, it is critical to balance both accuracy and efficiency in workload forecasting. In this paper, we propose XELASTIC, which offers three key improvements over the conference version. First, we redesigned the aggregation and disaggregation layers using GCNs to capture more complex relationships among workload series. Second, we introduced a supervised contrastive loss to enhance robustness against outliers, particularly for handling missing or abnormal data in real-world scenarios. Finally, we expanded the evaluation with additional baselines and larger datasets. Extensive experiments on realistic edge workload datasets collected from China’s largest edge service provider (Alibaba ENS) show that XELASTIC outperforms state-of-the-art methods, decreases time consumption, and reduces communication costs.
{"title":"A Collaborative Cloud-Edge Approach for Robust Edge Workload Forecasting","authors":"Yanan Li;Penghong Zhao;Xiao Ma;Haitao Yuan;Zhe Fu;Mengwei Xu;Shangguang Wang","doi":"10.1109/TMC.2024.3502683","DOIUrl":"https://doi.org/10.1109/TMC.2024.3502683","url":null,"abstract":"With the rapid development of edge computing in the post-COVID19 pandemic period, precise workload forecasting is considered the basis for making full use of the edge-limited resources, and both edge service providers (ESPs) and edge service consumers (ESCs) can benefit significantly from it. Existing paradigms of workload forecasting (i.e., edge-only or cloud-only) are improper, due to failing to consider the inter-site correlations and might suffer from significant data transmission delays. With the increasing adoption of edge platforms by web services, it is critical to balance both accuracy and efficiency in workload forecasting. In this paper, we propose XELASTIC, which offers three key improvements over the conference version. First, we redesigned the aggregation and disaggregation layers using GCNs to capture more complex relationships among workload series. Second, we introduced a supervised contrastive loss to enhance robustness against outliers, particularly for handling missing or abnormal data in real-world scenarios. Finally, we expanded the evaluation with additional baselines and larger datasets. Extensive experiments on realistic edge workload datasets collected from China’s largest edge service provider (Alibaba ENS) show that XELASTIC outperforms state-of-the-art methods, decreases time consumption, and reduces communication costs.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2861-2875"},"PeriodicalIF":7.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563920","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 : 2024-11-20DOI: 10.1109/TMC.2024.3502668
Kaikai Deng;Dong Zhao;Wenxin Zheng;Yue Ling;Kangwen Yin;Huadong Ma
Millimeter wave radar is gaining traction recently as a promising modality for enabling pervasive and privacy-preserving gesture recognition. However, the lack of rich and fine-grained radar datasets hinders progress in developing generalized deep learning models for gesture recognition across various user postures (e.g., standing, sitting), positions, and scenes. To remedy this, we resort to designing a software pipeline that exploits wealthy 2D videos to generate realistic radar data, but it needs to address the challenge of simulating diversified and fine-grained reflection properties of user gestures. To this end, we design G3R with three key components: i) a gesture reflection point generator expands the arm's skeleton points to form human reflection points; ii) a signal simulation model simulates the multipath reflection and attenuation of radar signals to output the human intensity map; iii) an encoder-decoder model combines a sampling module and a fitting module to address the differences in number and distribution of points between generated and real-world radar data for generating realistic radar data. We implement and evaluate G3R using 2D videos from public data sources and self-collected real-world radar data, demonstrating its superiority over other state-of-the-art approaches for gesture recognition.
{"title":"G3R: Generating Rich and Fine-Grained mmWave Radar Data From 2D Videos for Generalized Gesture Recognition","authors":"Kaikai Deng;Dong Zhao;Wenxin Zheng;Yue Ling;Kangwen Yin;Huadong Ma","doi":"10.1109/TMC.2024.3502668","DOIUrl":"https://doi.org/10.1109/TMC.2024.3502668","url":null,"abstract":"Millimeter wave radar is gaining traction recently as a promising modality for enabling pervasive and privacy-preserving gesture recognition. However, the lack of rich and fine-grained radar datasets hinders progress in developing generalized deep learning models for gesture recognition across various user postures (e.g., standing, sitting), positions, and scenes. To remedy this, we resort to designing a software pipeline that exploits wealthy 2D videos to generate realistic radar data, but it needs to address the challenge of simulating diversified and fine-grained reflection properties of user gestures. To this end, we design <monospace>G<sup>3</sup>R</monospace> with three key components: i) a <italic>gesture reflection point generator</i> expands the arm's skeleton points to form human reflection points; ii) a <italic>signal simulation model</i> simulates the multipath reflection and attenuation of radar signals to output the human intensity map; iii) an <italic>encoder-decoder model</i> combines a <italic>sampling module</i> and a <italic>fitting module</i> to address the differences in number and distribution of points between generated and real-world radar data for generating realistic radar data. We implement and evaluate <monospace>G<sup>3</sup>R</monospace> using 2D videos from public data sources and self-collected real-world radar data, demonstrating its superiority over other state-of-the-art approaches for gesture recognition.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":"24 4","pages":"2917-2934"},"PeriodicalIF":7.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143564012","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：