Joint DNN partitioning and task offloading in mobile edge computing via deep reinforcement learning DOI Creative Commons
Jianbing Zhang,

MA Shu-fang,

Zexiao Yan

et al.

Journal of Cloud Computing Advances Systems and Applications, Journal Year: 2023, Volume and Issue: 12(1)

Published: Aug. 3, 2023

Abstract As Artificial Intelligence (AI) becomes increasingly prevalent, Deep Neural Networks (DNNs) have become a crucial tool for developing and advancing AI applications. Considering limited computing energy resources on mobile devices (MDs), it is challenge to perform compute-intensive DNN tasks MDs. To attack this challenge, edge (MEC) provides viable solution through partitioning task offloading. However, as the communication conditions between different change over time, must also synchronously. This dynamic process, which aggravates complexity of partitioning. In paper, we delve into issue jointly optimizing delay offloading in MEC scenario where each MD server adopt pre-trained DNNs inference. Taking advantage characteristics DNN, first propose strategy layered divide subtasks that can be either processed or offloaded computation. Then, formulate trade-off joint optimization problem, further represented Markov decision process (MDP). solve this, design (DPTO) algorithm utilizing deep reinforcement learning (DRL), enables MDs make optimal decisions. Finally, experimental results demonstrate our outperforms existing non-DRL DRL algorithms with respect processing consumption, applied types.

Language: Английский

QoS-Aware Computation Offloading in LEO Satellite Edge Computing for IoT: A Game-Theoretical Approach DOI
Ying Chen, Jintao Hu, Jie Zhao

et al.

Chinese Journal of Electronics, Journal Year: 2024, Volume and Issue: 33(4), P. 875 - 885

Published: July 1, 2024

Language: Английский

Citations

27
Trustworthy and efficient project scheduling in IIoT based on smart contracts and edge computing DOI Creative Commons
Peng Liu, Xinglong Wu,

Yanjun Peng

et al.

Journal of Cloud Computing Advances Systems and Applications, Journal Year: 2025, Volume and Issue: 14(1)

Published: Jan. 11, 2025

To facilitate flexible manufacturing, modern industries have incorporated numerous modular operations such as multi-robot services which can be expediently arranged or offloaded to other production resources. However, complex manufacturing projects often consist of multiple tasks with fixed sequences, posing a significant challenge for smart factories in efficiently scheduling limited robot resources complete specific tasks. Additionally, when span across factories, ensuring faithful execution contracts becomes another challenge. In this paper, we propose modified combinatorial auction method combined blockchain and edge computing technologies organize project scheduling. Firstly, transform efficient resource into resource-constrained multi-project problem (RCPSP). Subsequently, the solution integrates random sampling (CA-RS) contracts. Alongside security analysis, simulations are conducted using real data sets. The results indicate that suggested CA-RS approach significantly enhances efficiency arrangement within industrial Internet Things compared baseline algorithms.

Language: Английский

Citations

1
A comprehensive review on internet of things task offloading in multi-access edge computing DOI Creative Commons

Wang Dayong,

Kamalrulnizam Bin Abu Bakar,

Babangida Isyaku

et al.

Heliyon, Journal Year: 2024, Volume and Issue: 10(9), P. e29916 - e29916

Published: April 22, 2024

With the rapid development of Internet Things (IoT) technology, Terminal Devices (TDs) are more inclined to offload computing tasks higher-performance servers, thereby solving problems insufficient capacity and battery consumption TD. The emergence Multi-access Edge Computing (MEC) technology provides new opportunities for IoT task offloading. It allows TDs access networks through multiple communication technologies supports mobility terminal devices. Review studies on offloading MEC have been extensive, but none them focus in MEC. To fill this gap, paper a comprehensive in-depth understanding algorithms mechanisms network. For each paper, main solved by mechanism, technical classification, evaluation methods, supported parameters extracted analyzed. Furthermore, shortcomings current research future trends discussed. This review will help potential researchers quickly understand panorama approaches find appropriate paths.

Language: Английский

Citations

6
Predicting the total Unified Parkinson’s Disease Rating Scale (UPDRS) based on ML techniques and cloud-based update DOI Creative Commons
Sahand Hamzehei, Omid Akbarzadeh,

Hani Attar

et al.

Journal of Cloud Computing Advances Systems and Applications, Journal Year: 2023, Volume and Issue: 12(1)

Published: Jan. 21, 2023

Abstract Nowadays, smart health technologies are used in different life and environmental areas, such as life, healthcare, cognitive cities, social systems. Intelligent, reliable, ubiquitous healthcare systems a part of the modern developing technology that should be more seriously considered. Data collection through ways, Internet things (IoT)-assisted sensors, enables physicians to predict, prevent treat diseases. Machine Learning (ML) algorithms may lead higher accuracy medical diagnosis/prognosis based on data provided by sensors help tracking symptom significance treatment steps. In this study, we applied four ML methods Parkinson’s disease assess methods’ performance identify essential features predict total Unified Rating Scale (UPDRS). Since accessibility high-performance decision-making so vital for updating supporting IoT nodes (e.g., wearable sensors), all is stored, updated rule-based, protected cloud. Moreover, assigning computational equipment memory use, cloud computing makes it possible reduce time complexity training phase cases want create complete structure cloud/edge architecture. situation, investigate approaches with varying iterations without concern system configuration, temporal complexity, real-time performance. Analyzing coefficient determination Mean Square Error (MSE) reveals outcomes mostly at an acceptable level. algorithm’s estimated weight indicates Motor UPDRS most significant predictor Total UPDRS.

Language: Английский

Citations

16
Reliable Integration of Neural Network and Internet of Things for Forecasting, Controlling, and Monitoring of Experimental Building Management System DOI Open Access
Mohamed El-Sayed M. Essa,

Ahmed M. El-shafeey,

Amna Hassan Omar

et al.

Sustainability, Journal Year: 2023, Volume and Issue: 15(3), P. 2168 - 2168

Published: Jan. 24, 2023

In this paper, Internet of Things (IoT) and artificial intelligence (AI) are employed to solve the issue energy consumption in a case study an education laboratory. IoT enables deployment AI approaches establish smart systems manage sensor signals between different equipment based on decisions. As result, paper introduces design investigation experimental building management system (BMS)-based approach monitor status sensors control operation loads reduce consumption. The proposed BMS is built integration programmable logic controller (PLC), Node MCU ESP8266, Arduino Mega 2560 perform roles transferring processing data as well decision-making. employs variety sensors, including DHT11 sensor, IR smoke ultrasonic sensor. collected from temperature used build neural network (ANN) model forecast inside platform created by ThingSpeak platform, Bylink dashboard, mobile application. results show that can publish platforms. addition, demonstrate air-conditioning, lighting, firefighting, ventilation could be optimally monitored managed for with architectural design. Furthermore, prove ANN distinct forecasting process data.

Language: Английский

Citations

14
Dynamic performance modeling framework for QoS-aware 5G-based IoT-edge systems DOI
Sujit Bebortta, Bibudhendu Pati, Chhabi Rani Panigrahi

et al.

International Journal of Information Technology, Journal Year: 2024, Volume and Issue: 16(4), P. 2149 - 2160

Published: Feb. 24, 2024

Language: Английский

Citations

5
Task offloading in hybrid-decision-based multi-cloud computing network: a cooperative multi-agent deep reinforcement learning DOI Creative Commons
Juan Chen, Peng Chen, Xianhua Niu

et al.

Journal of Cloud Computing Advances Systems and Applications, Journal Year: 2022, Volume and Issue: 11(1)

Published: Dec. 8, 2022

Abstract Multi-cloud computing is becoming a promising paradigm to provide abundant computation resources for Internet-of-Things (IoT) devices. For multi-device multi-cloud network, the real-time requirements, frequently varied wireless channel gains and changeable network scale, make system more dynamic. It critical satisfy dynamic nature of with different constraints IoT devices in environment. In this paper, we establish continuous-discrete hybrid decision offloading model, each device should learn coordinated actions, including cloud server selection, ratio local capacity. Therefore, both coordination among are challenging. To end, first develop probabilistic method relax discrete action (e.g. selection) continuous set. Then, by leveraging centralized training distributed execution strategy, design cooperative multi-agent deep reinforcement learning (CMADRL) based framework minimize total cost terms energy consumption renting charge servers. Each acts as an agent, which not only learns efficient decentralized policies, but also relieves devices’ pressure. Experimental results demonstrate that proposed CMADRL could efficiently polices at device, significantly outperform four state-of-the-art DRL agents two heuristic algorithms lower cost.

Language: Английский

Citations

20
Editorial: AI and IoT applications of smart buildings and smart environment design, construction and maintenance DOI
Ke Yan, Xiaokang Zhou, Bin Yang

et al.

Building and Environment, Journal Year: 2022, Volume and Issue: 229, P. 109968 - 109968

Published: Dec. 30, 2022

Language: Английский

Citations

19
A systematic and comprehensive review and investigation of intelligent IoT-based healthcare systems in rural societies and governments DOI
Yisu Ge, Guodao Zhang, Maytham N. Meqdad

et al.

Artificial Intelligence in Medicine, Journal Year: 2023, Volume and Issue: 146, P. 102702 - 102702

Published: Nov. 2, 2023

Language: Английский

Citations

11
Passive Infrared Sensor-Based Occupancy Monitoring in Smart Buildings: A Review of Methodologies and Machine Learning Approaches DOI Creative Commons
Azad Shokrollahi, Jan Persson, Reza Malekian

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(5), P. 1533 - 1533

Published: Feb. 27, 2024

Buildings are rapidly becoming more digitized, largely due to developments in the internet of things (IoT). This provides both opportunities and challenges. One central challenges process digitizing buildings is ability monitor these buildings’ status effectively. monitoring essential for services that rely on information about presence activities individuals within different areas buildings. Occupancy (including people counting, occupancy detection, location tracking, activity detection) plays a vital role management smart In this article, we primarily focus use passive infrared (PIR) sensors gathering information. PIR among most widely used purpose their consideration privacy concerns, cost-effectiveness, low processing complexity compared other sensors. Despite numerous literature reviews field information, there currently no review dedicated derived specifically from Therefore, analyzes articles explore application obtaining It comprehensive sensor technology 2015 2023, focusing applications localization (tracking location). consolidates findings have explored enhanced capabilities interconnected domains. thoroughly examines various techniques, machine learning algorithms, configurations indoor building environments, emphasizing not only data aspects but also advantages, limitations, efficacy producing accurate These crucial improving systems terms energy efficiency, security, user comfort, operational aspects. The article seeks offer thorough analysis present state potential future advancements efficiently understanding by classifying analyzing improvements

Language: Английский

Citations

4