Research

18 pages, 23085 KiB

Open AccessArticle

Reinforcement Learning Based Topology Control for UAV Networks

by Taehoon Yoo, Sangmin Lee, Kyeonghyun Yoo and Hwangnam Kim

Sensors 2023, 23(2), 921; https://doi.org/10.3390/s23020921 - 13 Jan 2023

Cited by 7 | Viewed by 2200

The recent development of unmanned aerial vehicle (UAV) technology has shown the possibility of using UAVs in many research and industrial fields. One of them is for UAVs moving in swarms to provide wireless networks in environments where there is no network infrastructure. [...] Read more.

The recent development of unmanned aerial vehicle (UAV) technology has shown the possibility of using UAVs in many research and industrial fields. One of them is for UAVs moving in swarms to provide wireless networks in environments where there is no network infrastructure. Although this method has the advantage of being able to provide a network quickly and at a low cost, it may cause scalability problems in multi-hop connectivity and UAV control when trying to cover a large area. Therefore, as more UAVs are used to form drone networks, the problem of efficiently controlling the network topology must be solved. To solve this problem, we propose a topology control system for drone networks, which analyzes relative positions among UAVs within a swarm, then optimizes connectivity among them in perspective of both interference and energy consumption, and finally reshapes a logical structure of drone networks by choosing neighbors per UAV and mapping data flows over them. The most important function in the scheme is the connectivity optimization because it should be adaptively conducted according to the dynamically changing complex network conditions, which includes network characteristics such as user density and UAV characteristics such as power consumption. Since neither a simple mathematical framework nor a network simulation tool for optimization can be a solution, we need to resort to reinforcement learning, specifically DDPG, with which each UAV can adjust its connectivity to other drones. In addition, the proposed system minimizes the learning time by flexibly changing the number of steps used for parameter learning according to the deployment of new UAVs. The performance of the proposed system was verified through simulation experiments and theoretical analysis on various topologies consisting of multiple UAVs. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

19 pages, 6237 KiB

Open AccessArticle

Distributed Raman Spectrum Data Augmentation System Using Federated Learning with Deep Generative Models

by Yaeran Kim and Woonghee Lee

Sensors 2022, 22(24), 9900; https://doi.org/10.3390/s22249900 - 16 Dec 2022

Cited by 8 | Viewed by 2175

Abstract

Chemical agents are one of the major threats to soldiers in modern warfare, so it is so important to detect chemical agents rapidly and accurately on battlefields. Raman spectroscopy-based detectors are widely used but have many limitations. The Raman spectrum changes unpredictably due [...] Read more.

Chemical agents are one of the major threats to soldiers in modern warfare, so it is so important to detect chemical agents rapidly and accurately on battlefields. Raman spectroscopy-based detectors are widely used but have many limitations. The Raman spectrum changes unpredictably due to various environmental factors, and it is hard for detectors to make appropriate judgments about new chemical substances without prior information. Thus, the existing detectors with inflexible techniques based on determined rules cannot deal with such problems flexibly and reactively. Artificial intelligence (AI)-based detection techniques can be good alternatives to the existing techniques for chemical agent detection. To build AI-based detection systems, sufficient amounts of data for training are required, but it is not easy to produce and handle fatal chemical agents, which causes difficulty in securing data in advance. To overcome the limitations, in this paper, we propose the distributed Raman spectrum data augmentation system that leverages federated learning (FL) with deep generative models, such as generative adversarial network (GAN) and autoencoder. Furthermore, the proposed system utilizes various additional techniques in combination to generate a large number of Raman spectrum data with reality along with diversity. We implemented the proposed system and conducted diverse experiments to evaluate the system. The evaluation results validated that the proposed system can train the models more quickly through cooperation among decentralized troops without exchanging raw data and generate realistic Raman spectrum data well. Moreover, we confirmed that the classification model on the proposed system performed learning much faster and outperformed the existing systems. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

15 pages, 5711 KiB

Open AccessArticle

Efficient Massive Computing for Deformable Volume Data Using Revised Parallel Resampling

by Chailim Park and Heewon Kye

Sensors 2022, 22(16), 6276; https://doi.org/10.3390/s22166276 - 20 Aug 2022

Cited by 1 | Viewed by 1302

Abstract

In this paper, we propose an improved parallel resampling technique. Parallel resampling is a deformable object generation method based on volume data applied to medical simulations. Existing parallel resampling is not suitable for massive computing, because the number of samplings is high and [...] Read more.

In this paper, we propose an improved parallel resampling technique. Parallel resampling is a deformable object generation method based on volume data applied to medical simulations. Existing parallel resampling is not suitable for massive computing, because the number of samplings is high and floating-point precision problems may occur. This study addresses these problems to obtain improved user latency when performing medical simulations. Specifically, instead of interpolating values after volume sampling, the efficiency is improved by performing volume sampling after coordinate interpolation. Next, the floating-point error in the calculation of the sampling position is described, and the advantage of barycentric interpolation using a reference point is discussed. The experimental results showed a significant improvement over the existing method. Volume data comprising more than 600 images used in clinical practice were deformed and rendered at interactive speed. In an Internet of Everything environment, medical imaging systems are an important application, and simulation image generation is also valuable in the overall system. Through the proposed method, the performance of the whole system can be improved. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

22 pages, 2787 KiB

Open AccessArticle

Time-Constrained Adversarial Defense in IoT Edge Devices through Kernel Tensor Decomposition and Multi-DNN Scheduling

by Myungsun Kim and Sanghyun Joo

Sensors 2022, 22(15), 5896; https://doi.org/10.3390/s22155896 - 07 Aug 2022

Viewed by 1361

Abstract

The development of deep learning technology has resulted in great contributions in many artificial intelligence services, but adversarial attack techniques on deep learning models are also becoming more diverse and sophisticated. IoT edge devices take cloud-independent on-device DNN (deep neural network) processing technology [...] Read more.

The development of deep learning technology has resulted in great contributions in many artificial intelligence services, but adversarial attack techniques on deep learning models are also becoming more diverse and sophisticated. IoT edge devices take cloud-independent on-device DNN (deep neural network) processing technology to exhibit a fast response time. However, if the computational complexity of the denoizer for adversarial noises is high, or if a single embedded GPU is shared by multiple DNN models, adversarial defense at the on-device level is bound to represent a long latency. To solve this problem, eDenoizer is proposed in this paper. First, it applies Tucker decomposition to reduce the computational amount required for convolutional kernel tensors in the denoizer. Second, eDenoizer effectively orchestrates both the denoizer and the model defended by the denoizer simultaneously. In addition, the priority of the CPU side can be projected onto the GPU which is completely priority-agnostic, so that the delay can be minimized when the denoizer and the defense target model are assigned a high priority. As a result of confirming through extensive experiments, the reduction of classification accuracy was very marginal, up to 1.78%, and the inference speed accompanied by adversarial defense was improved up to 51.72%. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

21 pages, 879 KiB

Open AccessArticle

Forecasting Obsolescence of Components by Using a Clustering-Based Hybrid Machine-Learning Algorithm

by Kyoung-Sook Moon, Hee Won Lee, Hee Jean Kim, Hongjoong Kim, Jeehoon Kang and Won Chul Paik

Sensors 2022, 22(9), 3244; https://doi.org/10.3390/s22093244 - 23 Apr 2022

Cited by 3 | Viewed by 1509

Abstract

Product obsolescence occurs in every production line in the industry as better-performance or cost-effective products become available. A proactive strategy for obsolescence allows firms to prepare for such events and reduces the manufacturing loss, which eventually leads to positive customer satisfaction. We propose [...] Read more.

Product obsolescence occurs in every production line in the industry as better-performance or cost-effective products become available. A proactive strategy for obsolescence allows firms to prepare for such events and reduces the manufacturing loss, which eventually leads to positive customer satisfaction. We propose a machine learning-based algorithm to forecast the obsolescence date of electronic diodes, which has a limitation on the amount of data available. The proposed algorithm overcomes these limitations in two ways. First, an unsupervised clustering algorithm is applied to group the data based on their similarity and build independent machine-learning models specialized for each group. Second, a hybrid method including several reliable techniques is constructed to improve the prediction accuracy and overcome the limitation of the lack of data. It is empirically confirmed that the prediction accuracy of the obsolescence date for the electrical component data is improved through the proposed clustering-based hybrid method. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

19 pages, 13222 KiB

Open AccessArticle

Federated Reinforcement Learning Based AANs with LEO Satellites and UAVs

by Seungho Yoo and Woonghee Lee

Sensors 2021, 21(23), 8111; https://doi.org/10.3390/s21238111 - 04 Dec 2021

Cited by 5 | Viewed by 2440

Abstract

Supported by the advances in rocket technology, companies like SpaceX and Amazon competitively have entered the satellite Internet business. These companies said that they could provide Internet service sufficiently to users using their communication resources. However, the Internet service might not be provided [...] Read more.

Supported by the advances in rocket technology, companies like SpaceX and Amazon competitively have entered the satellite Internet business. These companies said that they could provide Internet service sufficiently to users using their communication resources. However, the Internet service might not be provided in densely populated areas, as the satellites coverage is broad but its resource capacity is limited. To offload the traffic of the densely populated area, we present an adaptable aerial access network (AAN), composed of low-Earth orbit (LEO) satellites and federated reinforcement learning (FRL)-enabled unmanned aerial vehicles (UAVs). Using the proposed system, UAVs could operate with relatively low computation resources than centralized coverage management systems. Furthermore, by utilizing FRL, the system could continuously learn from various environments and perform better with the longer operation times. Based on our proposed design, we implemented FRL, constructed the UAV-aided AAN simulator, and evaluated the proposed system. Base on the evaluation result, we validated that the FRL enabled UAV-aided AAN could operate efficiently in densely populated areas where the satellites cannot provide sufficient Internet services, which improves network performances. In the evaluations, our proposed AAN system provided about 3.25 times more communication resources and had 5.1% lower latency than the satellite-only AAN. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

10 pages, 2214 KiB

Open AccessCommunication

Low-Complexity Transmit Power Control for Secure Communications in Wireless-Powered Cognitive Radio Networks

by Kisong Lee

Sensors 2021, 21(23), 7837; https://doi.org/10.3390/s21237837 - 25 Nov 2021

Cited by 1 | Viewed by 1126

Abstract

In this study, wireless-powered cognitive radio networks (WPCRNs) are considered, in which N sets of transmitters, receivers and energy-harvesting (EH) nodes in secondary networks share the same spectrum with primary users (PUs) and none of the EH nodes is allowed to decode information [...] Read more.

In this study, wireless-powered cognitive radio networks (WPCRNs) are considered, in which N sets of transmitters, receivers and energy-harvesting (EH) nodes in secondary networks share the same spectrum with primary users (PUs) and none of the EH nodes is allowed to decode information but can harvest energy from the signals. Given that the EH nodes are untrusted nodes from the point of view of information transfer, the eavesdropping of secret information can occur if they decide to eavesdrop on information instead of harvesting energy from the signals transmitted by secondary users (SUs). For secure communications in WPCRNs, we aim to find the optimal transmit powers of SUs that maximize the average secrecy rate of SUs while maintaining the interference to PUs below an allowable level, while guaranteeing the minimum EH requirement for each EH node. First, we derive an analytical expression for the transmit power via dual decomposition and propose a suboptimal transmit power control algorithm, which is implemented in an iterative manner with low complexity. The simulation results confirm that the proposed scheme outperforms the conventional distributed schemes by more than 10% in terms of the average secrecy rate and outage probability and can also considerably reduce the computation time compared with the optimal scheme. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

16 pages, 840 KiB

Open AccessArticle

Continuous Productivity Improvement Using IoE Data for Fault Monitoring: An Automotive Parts Production Line Case Study

by Yuchang Won, Seunghyeon Kim, Kyung-Joon Park and Yongsoon Eun

Sensors 2021, 21(21), 7366; https://doi.org/10.3390/s21217366 - 05 Nov 2021

Cited by 5 | Viewed by 2220

Abstract

This paper presents a case study of continuous productivity improvement of an automotive parts production line using Internet of Everything (IoE) data for fault monitoring. Continuous productivity improvement denotes an iterative process of analyzing and updating the production line configuration for productivity improvement [...] Read more.

This paper presents a case study of continuous productivity improvement of an automotive parts production line using Internet of Everything (IoE) data for fault monitoring. Continuous productivity improvement denotes an iterative process of analyzing and updating the production line configuration for productivity improvement based on measured data. Analysis for continuous improvement of a production system requires a set of data (machine uptime, downtime, cycle-time) that are not typically monitored by a conventional fault monitoring system. Although productivity improvement is a critical aspect for a manufacturing site, not many production systems are equipped with a dedicated data recording system towards continuous improvement. In this paper, we study the problem of how to derive the dataset required for continuous improvement from the measurement by a conventional fault monitoring system. In particular, we provide a case study of an automotive parts production line. Based on the data measured by the existing fault monitoring system, we model the production system and derive the dataset required for continuous improvement. Our approach provides the expected amount of improvement to operation managers in a numerical manner to help them make a decision on whether they should modify the line configuration or not. Full article

(This article belongs to the Special Issue Massive Learning and Computing for the Reliable Internet of Everything)

► Show Figures

Figure 1

Journal Menu

Journal Browser

Massive Learning and Computing for the Reliable Internet of Everything

Share This Special Issue

Special Issue Editors

Special Issue Information

Keywords

Published Papers (8 papers)

Research

Further Information

Guidelines

MDPI Initiatives

Follow MDPI