IEEE Transactions on Mobile Computing最新文献

筛选
英文 中文
A Two-Stage Deep Reinforcement Learning Framework for MEC-Enabled Adaptive 360-Degree Video Streaming 用于 MEC 自适应 360 度视频流的两阶段深度强化学习框架
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3443200
Suzhi Bi;Haoguo Chen;Xian Li;Shuoyao Wang;Yuan Wu;Liping Qian
{"title":"A Two-Stage Deep Reinforcement Learning Framework for MEC-Enabled Adaptive 360-Degree Video Streaming","authors":"Suzhi Bi;Haoguo Chen;Xian Li;Shuoyao Wang;Yuan Wu;Liping Qian","doi":"10.1109/TMC.2024.3443200","DOIUrl":"https://doi.org/10.1109/TMC.2024.3443200","url":null,"abstract":"The emerging multi-access edge computing (MEC) technology effectively enhances the wireless streaming performance of 360-degree videos. By connecting a user's head-mounted device (HMD) to a smart MEC platform, the edge server (ES) can efficiently perform adaptive tile-based video streaming to improve the user's viewing experience. Under constrained wireless channel capacity, the ES can predict the user's field of view (FoV) and transmit to the HMD high-resolution video tiles only within the predicted FoV. In practice, the video streaming performance is challenged by the random FoV prediction error and wireless channel fading effects. For this, we propose in this paper a novel two-stage adaptive 360-degree video streaming scheme that maximizes the user's quality of experience (QoE) to attain stable and high-resolution video playback. Specifically, we divide the video file into groups of pictures (GOPs) of fixed playback interval, where each GOP consists of a number of video frames. At the beginning of each GOP (i.e., the inter-GOP stage), the ES predicts the FoV of the next GOP and allocates an encoding bitrate for transmitting (precaching) the video tiles within the predicted FoV. Then, during the real-time video playback of the current GOP (i.e., the intra-GOP stage), the ES observes the user's true FoV of each frame and transmits the missing tiles to compensate for the FoV prediction errors. To maximize the user's QoE under random variations of FoV and wireless channel, we propose a double-agent deep reinforcement learning framework, where the two agents operate in different time scales to decide the bitrates of inter- and intra-GOP stages, respectively. Experiments based on real-world measurements show that the proposed scheme can effectively mitigate FoV prediction errors and maintain stable QoE performance under different scenarios, achieving over 22.1% higher QoE than some representative benchmark methods.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142598657","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}
引用次数: 0
VNF Scheduling and Sampling Rate Maximization in Energy Harvesting IoT Networks 能量收集物联网网络中的 VNF 调度和采样率最大化
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442809
Longji Zhang;Kwan-Wu Chin
{"title":"VNF Scheduling and Sampling Rate Maximization in Energy Harvesting IoT Networks","authors":"Longji Zhang;Kwan-Wu Chin","doi":"10.1109/TMC.2024.3442809","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442809","url":null,"abstract":"This paper studies virtual network function (VNF) scheduling in energy harvesting virtualized Internet of Things (IoT) networks. Unlike prior works, sensor devices leverage imprecise computation to vary their computational workload to conserve energy at the expense of computation quality. In this respect, an optimization problem of interest is to maximize the minimum VNF computation/execution quality. To this end, this paper presents the first mixed integer linear program (MILP) that optimizes i) the VNFs executed by each sensor device, ii) the computational resources allocated to VNFs, iii) sampling rate or amount of data supplied by sensor devices to VNFs, iv) the routing of samples to VNFs and forwarding of computation results, and v) link scheduling. In addition, this paper also proposes a heuristic, called sampling control and computation scheduling (SCACS), for large-scale networks. The simulation results show that SCACS reaches 81.66% of the optimal quality. In addition, the application completion rate when using SCACS is at most 39% higher than a benchmark that randomly selects nodes to sample targets and execute VNFs.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142598639","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}
引用次数: 0
RIC-SDA: A Reputation Incentive Committee-Based Secure Conditional Dual Authentication Scheme for VANETs RIC-SDA:基于声誉激励委员会的 VANET 安全条件双重认证方案
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442933
Ningbin Yang;Chunming Tang;Tianqi Zong;Zhikang Zeng;Zehui Xiong;Debiao He
{"title":"RIC-SDA: A Reputation Incentive Committee-Based Secure Conditional Dual Authentication Scheme for VANETs","authors":"Ningbin Yang;Chunming Tang;Tianqi Zong;Zhikang Zeng;Zehui Xiong;Debiao He","doi":"10.1109/TMC.2024.3442933","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442933","url":null,"abstract":"Vehicular ad hoc networks (VANETs) establish wireless connections among all vehicles, enabling seamless mobile communication. However, existing conditional privacy protection VANETs authentication schemes fail to address the issue of potential key-exposure and do not provide accelerated vehicle authentication. In this paper, we propose a reputation incentive committee-based secure conditional dual authentication scheme for VANETs called RIC-SDA. Our proposed scheme incorporates dual authentication of the consensus committee and vehicle-to-vehicle (V2V) communication. It enables the rapid provision of dynamic vehicle epoch-key from consensus committee authentication for V2V authentication through our designed reputation incentive mechanism. To mitigate the potential key-exposure problem, we introduce a novel concept of secure vehicle epoch communication, which means V2V authentication is valid for only one epoch blockchain unit time. The proposed scheme achieves lightweight computation and incurs minimal communication overheads, with the signature size being just 137 bytes. The RIC-SDA scheme supports fast batch verification. We prove that our proposed scheme is unforgeable security under random oracle and demonstrate its feasibility by implementing it in a test network based on Ethereum Sepolia. The results demonstrate that our RIC-SDA solution outperforms the existing state-of-the-art authentication VANET schemes regarding efficiency and communication costs.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142598647","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}
引用次数: 0
Efficient, Scalable, and Sustainable DNN Training on SoC-Clustered Edge Servers 在 SoC 集群边缘服务器上进行高效、可扩展和可持续的 DNN 训练
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442430
Mengwei Xu;Daliang Xu;Chiheng Lou;Li Zhang;Gang Huang;Xin Jin;Xuanzhe Liu
{"title":"Efficient, Scalable, and Sustainable DNN Training on SoC-Clustered Edge Servers","authors":"Mengwei Xu;Daliang Xu;Chiheng Lou;Li Zhang;Gang Huang;Xin Jin;Xuanzhe Liu","doi":"10.1109/TMC.2024.3442430","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442430","url":null,"abstract":"In the realm of industrial edge computing, a novel server architecture known as SoC-Cluster, characterized by its aggregation of numerous mobile systems-on-chips (SoCs), has emerged as a promising solution owing to its enhanced energy efficiency and seamless integration with prevalent mobile applications. Despite its advantages, the utilization of SoC-Cluster servers remains unsatisfactory, primarily attributed to the tidal patterns of user-initiated workloads. To address such inefficiency, we introduce \u0000<monospace>SoCFlow+</monospace>\u0000, a pioneering framework designed to facilitate the co-location of deep learning training tasks on SoC-Cluster servers, thereby optimizing resource utilization. \u0000<monospace>SoCFlow+</monospace>\u0000 incorporates three novel techniques tailored to mitigate the inherent limitations of commercial SoC-Cluster servers. First, it employs group-wise parallelism complemented by delayed aggregation, a strategy engineered to enhance the training efficiency and scalability of deep learning models, effectively circumventing network bottlenecks. Second, it integrates a data-parallel mixed-precision training algorithm, optimized to exploit the heterogeneous processing capabilities inherent to mobile SoCs fully. Third, \u0000<monospace>SoCFlow+</monospace>\u0000 employs an underclocking-aware workload re-balanacing mechanism to tackle the training performance degradation caused by the thermal control of mobile SoCs. Through rigorous experimental validation, \u0000<monospace>SoCFlow+</monospace>\u0000 achieves a convergence speedup ranging from 1.6× to 740× across 32 SoCs, compared to conventional benchmarks. Furthermore, when juxtaposed with commodity GPU servers (e.g., NVIDIA V100) under identical power constraints, \u0000<monospace>SoCFlow+</monospace>\u0000 not only exhibits comparable training speed but also achieves a remarkable reduction in energy consumption by a factor of 2.31× to 10.23×, all while preserving convergence accuracy.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142598666","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}
引用次数: 0
Ubiquitous Indoor Mapping Using Mobile Radio Tomography 利用移动无线电层析成像技术进行无所不在的室内测绘
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442439
Amartya Basu;Ayon Chakraborty;Kush Jajal
{"title":"Ubiquitous Indoor Mapping Using Mobile Radio Tomography","authors":"Amartya Basu;Ayon Chakraborty;Kush Jajal","doi":"10.1109/TMC.2024.3442439","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442439","url":null,"abstract":"The demand for real-time and accurate mapping is ubiquitous, particularly in complex indoor settings. While SLAM-based methods are popular, Radio Tomographic Imaging (RTI) offers an essential set of advantages, including mapping inaccessible or enclosed spaces, shorter scanning trajectories, or even identifying material properties of structures on the map. However, existing RTI systems typically depend on pre-deployed, precisely calibrated infrastructure with ample computing power, making it challenging to deploy in a ubiquitous setting. We design \u0000<sc>UbiqMap</small>\u0000, a lightweight RTI-based end-to-end system capable of mapping indoor spaces in real-time, with minimal to zero reliance over pre-deployed infrastructure. We evaluate the performance of \u0000<sc>UbiqMap</small>\u0000 in various scenarios, including two real deployments - a moderately complex residential apartment (800 sq. ft) and a large building foyer area (3000 sq. ft) and a few simulated scenarios. We demonstrate how \u0000<sc>UbiqMap</small>\u0000 can benefit over traditional SLAM-based techniques in specific contexts and advocate the fusion of RTI methods with SLAM to improve future mapping technologies. Overall, \u0000<sc>UbiqMap</small>\u0000 improves the quality of the estimated map by 30%–40% over the state-of-the-art with equivalent resource availability.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595921","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}
引用次数: 0
FLuMe: Understanding Differential Spectrum Mobility Features in High Resolution FLuMe:了解高分辨率下的差异频谱移动特征
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442151
Rui Zou;Wenye Wang
{"title":"FLuMe: Understanding Differential Spectrum Mobility Features in High Resolution","authors":"Rui Zou;Wenye Wang","doi":"10.1109/TMC.2024.3442151","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442151","url":null,"abstract":"Existing measurements and modeling of radio spectrum usage have shown that exclusive access leads to low efficiency. Thus, the next generation of wireless networks is adopting new paradigms of spectrum sharing and coexistence among heterogeneous networks. However, two significant limitations in current spectrum tenancy models hinder the development of essential functions in nonexclusive spectrum access. First, these models rely on data with much coarser resolutions than those required for wireless scheduling, rendering them ineffective for spectrum prediction or characterizing spectrum access behavior in a wireless coexistence setting. Second, due to a lack of detailed data, current models cannot describe the access dynamics of individual users, leading to unjustified adoption of simplistic traffic models, such as the on/off model and the M/G/1 queue, in spectrum access algorithm research. To address these limitations, we propose the Frame-Level spectrum Model (FLuMe), a data-driven model that characterizes individual spectrum usage based on high-resolution data. This lightweight model tracks the spectrum tenancy movements of individual users using four variables. The proposed model is applied to high-resolution LTE spectrum tenancy data, from which model parameters are extracted. Comprehensive validations demonstrate the goodness-of-fit of the model and its applicability to spectrum prediction.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595839","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}
引用次数: 0
Cell-Less Offloading of Distributed Learning Tasks in Multi-Access Edge Computing 多接入边缘计算中分布式学习任务的无单元卸载
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-13 DOI: 10.1109/TMC.2024.3442242
Pengchao Han;Bo Liu;Yejun Liu;Lei Guo
{"title":"Cell-Less Offloading of Distributed Learning Tasks in Multi-Access Edge Computing","authors":"Pengchao Han;Bo Liu;Yejun Liu;Lei Guo","doi":"10.1109/TMC.2024.3442242","DOIUrl":"https://doi.org/10.1109/TMC.2024.3442242","url":null,"abstract":"Multi-access edge computing (MEC) is a powerful technology that facilitates the provision of services to 6G users with ultra-low latency and high reliability, particularly in supporting artificial intelligence (AI) applications that rely on distributed machine learning (DL). However, the mobility of users poses challenges in offloading DL tasks to the MEC networks while ensuring satisfactory delay and blocking rates. Task replication emerges as a promising technique for achieving a cell-less design for mobile users. Nevertheless, existing research overlooks the replication of DL tasks involving multiple subtasks and users, as well as the high resource cost of task replication. Towards this challenge, this paper investigates the Mobility-awarE mulTi-replicA (META) DL task offloading problem in MEC networks. First, we propose a hybrid resource allocation mechanism that allocates resources to a replica with high access probability in a static manner and dynamically allocates resources to replicas with low access probabilities. Then, we develop an access base station (BS) clustering algorithm for each user to determine the optimal number of replicas. Additionally, we propose the META DL task offloading algorithms with proved approximation ratios to minimize the overall resource cost. Through simulations based on generated and real-world mobile users, we demonstrate the effectiveness of our proposed algorithms.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142598635","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}
引用次数: 0
Privacy-Preserving and Secure Distributed Data Sharing Scheme for VANETs 面向 VANET 的隐私保护和安全分布式数据共享方案
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-12 DOI: 10.1109/TMC.2024.3441595
Li Wang;Hong Zhong;Jie Cui;Jing Zhang;Lu Wei;Irina Bolodurina;Debiao He
{"title":"Privacy-Preserving and Secure Distributed Data Sharing Scheme for VANETs","authors":"Li Wang;Hong Zhong;Jie Cui;Jing Zhang;Lu Wei;Irina Bolodurina;Debiao He","doi":"10.1109/TMC.2024.3441595","DOIUrl":"https://doi.org/10.1109/TMC.2024.3441595","url":null,"abstract":"Data sharing is one of the essential services of vehicular ad hoc networks (VANETs), which primarily requires data security and access control, and ciphertext-policy attribute-based encryption (CP-ABE) is a promising tool. However, data sharing schemes of distributed CP-ABE have concerns about the single-point performance bottleneck and privacy leakage. The factor for the former is that the authority manages a disjoint attribute set. The latter is because the user's identity and attributes are required to submit to authorities, which targets to bind this information to decryption keys for collusion-resistant. We propose a privacy-preserving distributed data sharing scheme for VANETs. This scheme introduces asymmetric group key agreement to distributed CP-ABE, which realizes that multiple authorities manage an attribute, and the user can obtain the attribute key bound with his identity from any authority in the group. To match up to the requirement of privacy-preserving, a key extract protocol provided user anonymity is proposed, which implements that attribute keys can be obtained without revealing the user's identity and attributes. Moreover, partial policy hiding is satisfied. Finally, we analyze and evaluate the proposed scheme, and the results indicate that our scheme is secure and efficient.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595898","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}
引用次数: 0
Graph Based RFID Grouping for Fast and Robust Inventory Tracking 基于图的 RFID 分组技术实现快速可靠的库存跟踪
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-12 DOI: 10.1109/TMC.2024.3439430
Meng Jin;Kexin Li;Xiaohua Tian;Xinbing Wang;Chenghu Zhou
{"title":"Graph Based RFID Grouping for Fast and Robust Inventory Tracking","authors":"Meng Jin;Kexin Li;Xiaohua Tian;Xinbing Wang;Chenghu Zhou","doi":"10.1109/TMC.2024.3439430","DOIUrl":"https://doi.org/10.1109/TMC.2024.3439430","url":null,"abstract":"This paper presents the design, implementation, and evaluation of TaGroup, a fast, fine-grained, and robust grouping technique for RFIDs. It can achieve a nearly 100% accuracy in distinguishing multiple groups of closely located RFIDs, within only a few seconds. It would benefit many inventory tracking applications, such as self-checkout in retails and packaging quality control in logistics. We make two technical innovations. First, we propose a novel method which can measure the channels between multiple pairs of commercial RFID tags simultaneously, and then estimate the proximity relations between them based on the channel information. Second, we introduce a spatio-temporal graph model which captures a full picture of proximity relations among all the tags, based on which TaGroup can perform a robust grouping of the tags. These two designs together boost the grouping speed and accuracy of TaGroup. Our experiments show that in grouping 120 tags into 4 closely located groups, TaGroup can achieve a nearly 100% accuracy, at the cost of only 2 seconds.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595922","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}
引用次数: 0
Elastic DNN Inference With Unpredictable Exit in Edge Computing 边缘计算中不可预测出口的弹性 DNN 推断
IF 7.7 2区 计算机科学
IEEE Transactions on Mobile Computing Pub Date : 2024-08-12 DOI: 10.1109/TMC.2024.3441946
Jiaming Huang;Yi Gao;Wei Dong
{"title":"Elastic DNN Inference With Unpredictable Exit in Edge Computing","authors":"Jiaming Huang;Yi Gao;Wei Dong","doi":"10.1109/TMC.2024.3441946","DOIUrl":"https://doi.org/10.1109/TMC.2024.3441946","url":null,"abstract":"Multi-exit neural networks have gained popularity in edge computing to leverage the computing power of diverse devices. However, real-time tasks in edge applications often face frequent unpredictable exits caused by power outages or high-priority preemptions, which have been largely overlooked by multi-exit models. To address this challenge, it is crucial to determine the appropriate exit point in the multi-exit model to ensure desirable results during unpredictable exits. In this paper, we propose EINet, a sample-wise planner for real-time multi-exit deep neural networks. EINet enables efficient Elastic Inference with unpredictable exits while ensuring best-effort accuracy on various edge platforms. Our approach involves partitioning a trained deep neural network into multiple blocks, each with its exit. Furthermore, EINet utilizes block-wise model profiles, which include accuracy and inference time information for each block. By leveraging these profiles, EINet dynamically determines the optimal exit plan for each sample during the inference process. We introduce Confidence Score Predictors to adapt to the unique characteristics of input samples and employ the Search Engine to efficiently find near-optimal plans for elastic inference. Extensive evaluations of EINet using multiple deep neural networks and datasets with unpredictable exits demonstrate its superior performance. EINet exhibits significant accuracy improvements: 0.13%–16.5% compared to static plans, 0.79%–4.1% compared to other dynamic plans, and over 50% compared to predictable inference in typical scenarios.","PeriodicalId":50389,"journal":{"name":"IEEE Transactions on Mobile Computing","volume":null,"pages":null},"PeriodicalIF":7.7,"publicationDate":"2024-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595896","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}
引用次数: 0
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
相关产品
×
本文献相关产品
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信