[go: up one dir, main page]
Next Issue
Volume 19, August-2
Previous Issue
Volume 19, July-2
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
7.3
3.4
Journals
Sensors
Volume 19
Issue 15
sensors-logo

Journal Menu

► Journal Menu

Journal Browser

► Journal Browser
share announcement

Sensors, Volume 19, Issue 15 (August-1 2019) – 198 articles

Cover Story (view full-size image): A burst images sequence of the Sentinel-1B raw dataset was acquired through the terrain observation with progressive scan (TOPS) interferometric wide mode, processed using the phase-preserving focusing algorithm of TOPS synthetic aperture radar (SAR) data developed at IREA-CNR, Italy. The investigated raw data, acquired on 1 January 2019, include the Deutsche Zentrum für Luft-und Raumfahrt (DLR) calibration site, for a total extension of about 250 Km × 200 Km. The proposed algorithm consists of a first interpolation stage of the TOPS raw data, which allows unfolding the azimuth spectra of the TOPS raw data. Subsequently, the interpolated signals are processed using the conventional phase-preserving SAR focusing methods that exploit frequency domain and spectral analyses algorithms, extensively used to efficiently process Stripmap and ScanSAR data. View this paper.
  • Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
  • You may sign up for e-mail alerts to receive table of contents of newly released issues.
  • PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.
Order results
Result details
Section
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
15 pages, 3593 KiB  
Article
A Novel Resource Allocation and Spectrum Defragmentation Mechanism for IoT-Based Big Data in Smart Cities
by Yuhuai Peng, Jiaying Wang, Aiping Tan and Jingjing Wu
Sensors 2019, 19(15), 3443; https://doi.org/10.3390/s19153443 - 6 Aug 2019
Cited by 5 | Viewed by 3745
Abstract
People’s demand for high-traffic applications and the need for Internet of Things (IoT) are enormous in smart cities. The amount of data generated by virtual reality, high-definition video, and other IoT applications is growing at an exponential rate that far exceeds our expectations, [...] Read more.
People’s demand for high-traffic applications and the need for Internet of Things (IoT) are enormous in smart cities. The amount of data generated by virtual reality, high-definition video, and other IoT applications is growing at an exponential rate that far exceeds our expectations, and the types of data are becoming more diverse. It has become critical to reliably accommodate IoT-based big data with reasonable resource allocation in optical backbone networks for smart cities. For the problem of reliable transmission and efficient resource allocation in optical backbone networks, a novel resource allocation and spectrum defragmentation mechanism for massive IoT traffic is presented in this paper. Firstly, a routing and spectrum allocation algorithm based on the distance-adaptive sharing protection mechanism (DASP) is proposed, to obtain sufficient protection and reduce the spectrum consumption. The DASP algorithm advocates applying different strategies to the resource allocation of the working and protection paths. Then, the protection path spectrum defragmentation algorithm based on OpenFlow is designed to improve spectrum utilization while providing shared protection for traffic demands. The lowest starting slot-index first (LSSF) algorithm is employed to remove and reconstruct the optical paths. Numerical results indicate that the proposal can effectively alleviate spectrum fragmentation and reduce the bandwidth-blocking probability by 44.68% compared with the traditional scheme. Full article
(This article belongs to the Special Issue Big Data Driven IoT for Smart Cities)
►▼ Show Figures

Figure 1

Figure 1
<p>SDN-based network architecture.</p> Full article ">Figure 2
<p>SDN-based system function module, OF-AG: Openflow Agent, OF-C: Openflow Controller, OF-Client: Openflow Client, LTD: Local Traffic Database, DF Agent: Defragmentation Agent, NMS: Network Management System, RCM: Resource Computation Module, TED: Traffic Engineering Database, RPM: Resource Provisioning Module, NAM: Network Abstraction Module.</p> Full article ">Figure 3
<p>An eight-node network.</p> Full article ">Figure 4
<p>Spectrum usage status without spectrum sharing.</p> Full article ">Figure 5
<p>Spectrum usage status when spectrum sharing.</p> Full article ">Figure 6
<p>Spectrum resource usage status after defragmentation.</p> Full article ">Figure 7
<p>Comparison of BBP for various RSA scenarios in the NSFNET network.</p> Full article ">Figure 8
<p>Comparison of BBP for various RSA scenarios in the USNET network.</p> Full article ">Figure 9
<p>BBP with LSSF algorithm in the NSFENT network.</p> Full article ">Figure 10
<p>BBP with LSSF algorithm in the USNET network.</p> Full article ">
17 pages, 1068 KiB  
Article
Iterative Trajectory Optimization for Physical-Layer Secure Buffer-Aided UAV Mobile Relaying
by Lingfeng Shen, Ning Wang, Xiang Ji, Xiaomin Mu and Lin Cai
Sensors 2019, 19(15), 3442; https://doi.org/10.3390/s19153442 - 6 Aug 2019
Cited by 17 | Viewed by 3804
Abstract
With the fast development of commercial unmanned aerial vehicle (UAV) technology, there are increasing research interests on UAV communications. In this work, the mobility and deployment flexibility of UAVs are exploited to form a buffer-aided relaying system assisting terrestrial communication that is blocked. [...] Read more.
With the fast development of commercial unmanned aerial vehicle (UAV) technology, there are increasing research interests on UAV communications. In this work, the mobility and deployment flexibility of UAVs are exploited to form a buffer-aided relaying system assisting terrestrial communication that is blocked. Optimal UAV trajectory design of the UAV-enabled mobile relaying system with a randomly located eavesdropper is investigated from the physical-layer security perspective to improve the overall secrecy rate. Based on the mobility of the UAV relay, a wireless channel model that changes with the trajectory and is exploited for improved secrecy is established. The secrecy rate is maximized by optimizing the discretized trajectory anchor points based on the information causality and UAV mobility constraints. However, the problem is non-convex and therefore difficult to solve. To make the problem tractable, we alternatively optimize the increments of the trajectory anchor points iteratively in a two-dimensional space and decompose the problem into progressive convex approximate problems through the iterative procedure. Convergence of the proposed iterative trajectory optimization technique is proved analytically by the squeeze principle. Simulation results show that finding the optimal trajectory by iteratively updating the displacements is effective and fast converging. It is also shown by the simulation results that the distribution of the eavesdropper location influences the security performance of the system. Specifically, an eavesdropper further away from the destination is beneficial to the system’s overall secrecy rate. Furthermore, it is observed that eavesdropper being further away from the destination also results in shorter trajectories, which implies it being energy-efficient as well. Full article
(This article belongs to the Special Issue Selected Papers from CyberC 2018)
►▼ Show Figures

Figure 1

Figure 1
<p>The UAV-enabled mobile relaying system model.</p> Full article ">Figure 2
<p>Convergence of the average secrecy rate performance with different maximum UAV speed values. The total flight time is <math display="inline"><semantics> <mrow> <mi>T</mi> <mo>=</mo> <mn>80</mn> </mrow> </semantics></math> s. The proposed algorithm exhibits fast convergence property in all the scenarios examined. Higher maximum UAV speed results in higher average secrecy rate performance.</p> Full article ">Figure 3
<p>Average secrecy rate performance for different distribution boundaries of the eavesdropper location with maximum UAV speed <math display="inline"><semantics> <mrow> <mi>v</mi> <mo>=</mo> <mn>20</mn> </mrow> </semantics></math> m/s and total flight time <math display="inline"><semantics> <mrow> <mi>T</mi> <mo>=</mo> <mn>80</mn> </mrow> </semantics></math> s. The eavesdropper located further away from the destination is shown to be more favorable to the overall average secrecy rate performance.</p> Full article ">Figure 4
<p>The iterative updates of the UAV trajectory with maximum UAV speed <math display="inline"><semantics> <mrow> <mi>v</mi> <mo>=</mo> <mn>16</mn> </mrow> </semantics></math> m/s and total flight time <math display="inline"><semantics> <mrow> <mi>T</mi> <mo>=</mo> <mn>80</mn> </mrow> </semantics></math> s. The eavesdropper location is uniformly distributed between <math display="inline"><semantics> <mrow> <mo>[</mo> <mn>300</mn> <mo>,</mo> <mn>500</mn> <mo>]</mo> </mrow> </semantics></math> on the <math display="inline"><semantics> <msub> <mi>d</mi> <mi>x</mi> </msub> </semantics></math> axis. The UAV’s trajectory converges in about 10 iterations.</p> Full article ">Figure 5
<p>The optimized trajectories for different eavesdropper locations with maximum UAV speed <math display="inline"><semantics> <mrow> <mi>v</mi> <mo>=</mo> <mn>16</mn> </mrow> </semantics></math> m/s and total flight <math display="inline"><semantics> <mrow> <mi>T</mi> <mo>=</mo> <mn>80</mn> </mrow> </semantics></math> s. It is observed that when the eavesdropper location is further away from the destination, the UAV’s optimized trajectory has a shorter total flight distance, which is both spectrum-efficient and energy-efficient.</p> Full article ">
16 pages, 2339 KiB  
Article
Simultaneous Measurement of Ear Canal Movement, Electromyography of the Masseter Muscle and Occlusal Force for Earphone-Type Occlusal Force Estimation Device Development
by Mami Kurosawa, Kazuhiro Taniguchi, Hideya Momose, Masao Sakaguchi, Masayoshi Kamijo and Atsushi Nishikawa
Sensors 2019, 19(15), 3441; https://doi.org/10.3390/s19153441 - 6 Aug 2019
Cited by 9 | Viewed by 4467
Abstract
We intend to develop earphone-type wearable devices to measure occlusal force by measuring ear canal movement using an ear sensor that we developed. The proposed device can measure occlusal force during eating. In this work, we simultaneously measured the ear canal movement (ear [...] Read more.
We intend to develop earphone-type wearable devices to measure occlusal force by measuring ear canal movement using an ear sensor that we developed. The proposed device can measure occlusal force during eating. In this work, we simultaneously measured the ear canal movement (ear sensor value), the surface electromyography (EMG) of the masseter muscle and the occlusal force six times from five subjects as a basic study toward occlusal force meter development. Using the results, we investigated the correlation coefficient between the ear sensor value and the occlusal force, and the partial correlation coefficient between ear sensor values. Additionally, we investigated the average of the partial correlation coefficient and the absolute value of the average for each subject. The absolute value results indicated strong correlation, with correlation coefficients exceeding 0.9514 for all subjects. The subjects showed a lowest partial correlation coefficient of 0.6161 and a highest value of 0.8286. This was also indicative of correlation. We then estimated the occlusal force via a single regression analysis for each subject. Evaluation of the proposed method via the cross-validation method indicated that the root-mean-square error when comparing actual values with estimates for the five subjects ranged from 0.0338 to 0.0969. Full article
(This article belongs to the Special Issue Wearable Sensors and Devices for Healthcare Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Experimental system. In this system, analog signals ranging from 0 V to 3.3 V measured using occlusal force meter, electromyography and the ear sensor to measure the movement of the ear canal are converted into digital signals by the analog-to-digital converter at a sampling frequency of 100 Hz with 12 bit resolution; the digital signals are then recorded together with timestamps in a storage device.</p> Full article ">Figure 2
<p>Principle of ear canal movement measurement using ear sensor. Occlusion is performed by the temporalis and masticatory muscles, including the masseter muscle and the temporomandibular joint. Occlusion causes a change in the ear canal shape near the masticatory muscles and the temporomandibular joint. The ear sensor measures this shape change in the ear canal during occlusion optically and noninvasively. A small photosensor is attached to the ear sensor. This photosensor houses a light-emitting diode (LED) with an emission wavelength of 940 nm and a phototransistor, as illustrated in <a href="#sensors-19-03441-f001" class="html-fig">Figure 1</a>. The ear sensor irradiates the skin of the ear canal with infrared light, and the reflected light is then received by the phototransistor to measure the change in the ear canal shape.</p> Full article ">Figure 3
<p>Appearance of the GM-10 occlusal force meter. This occlusal force meter is constructed continuously of an intraoral insertion part and a gripping part; 88 mm of the total length is the intraoral insertion part (on the left side in the figure) and the remaining 101 mm is the gripping part (on the right side in the figure). During measurements, the disposable resin-made cover is placed on the intraoral insertion part in advance. The subject then holds the gripping part using a single hand and the sensor measures the occlusal force when the subject chews the tip (i.e., the sensing area) of the intraoral insertion part.</p> Full article ">Figure 4
<p>Measured results for subject A. The graph shows the measured results for which the correlation coefficient between the ear sensor value and the occlusal force was the highest among the six runs of subject A.</p> Full article ">Figure 5
<p>Measurement results for ear sensor and occlusal force over the first through sixth runs for subject A. Here, the horizontal axis represents the ear sensor-measured value, while the vertical axis represents the measured occlusal force value.</p> Full article ">
16 pages, 3486 KiB  
Article
Artificial Intelligence Assisted Heating Ventilation and Air Conditioning Control and the Unmet Demand for Sensors: Part 2. Prior Information Notice (PIN) Sensor Design and Simulation Results
by Chin-Chi Cheng and Dasheng Lee
Sensors 2019, 19(15), 3440; https://doi.org/10.3390/s19153440 - 6 Aug 2019
Cited by 5 | Viewed by 4521
Abstract
The study continues the theoretical derivation from Part 1, and the experiment is carried out at a bus station equipped with six water-cooled chillers. Between 2012 and 2017, historical data collected from temperature and humidity sensors, as well as the energy consumption data, [...] Read more.
The study continues the theoretical derivation from Part 1, and the experiment is carried out at a bus station equipped with six water-cooled chillers. Between 2012 and 2017, historical data collected from temperature and humidity sensors, as well as the energy consumption data, were used to build artificial intelligence (AI) assisted heating ventilation and air conditioning (HVAC) control models. The AI control system, in conjunction with a specifically designed prior information notice (PIN) sensor, was used to improve the prediction accuracy. This data collected between 2012 and 2016 was used for AI training and PIN sensor testing. During the hottest week of 2017 in Taiwan, the PIN sensor was used to conduct temperature and humidity data predictions. A model-based predictive control was developed to obtain air conditioning energy consumption data. The comparative results between the predictive and actual data showed that the temperature and humidity prediction accuracies were between 95.5 and 96.6%, respectively. Additionally, energy savings amounting to 39.8% were achieved compared to the theoretical estimates of 44.6%, a difference of less than 5%. These results show that the experimental model supports the theoretical estimations. In the future, a PIN sensor will be installed in a chiller to further verify the energy savings of the AI assisted HVAC control. Full article
(This article belongs to the Section Physical Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>A real test site for presenting artificial intelligence (AI)-assisted heating ventilation and air conditioning (HVAC) control performances. (<b>a</b>) The case study facility; (<b>b</b>) the chiller system including 6 chillers installed in the building; (<b>c</b>) a power meter for measuring the energy consumption; (<b>d</b>) a networked temperature/humidity sensor module for detecting the necessary parameters for the HVAC control.</p> Full article ">Figure 1 Cont.
<p>A real test site for presenting artificial intelligence (AI)-assisted heating ventilation and air conditioning (HVAC) control performances. (<b>a</b>) The case study facility; (<b>b</b>) the chiller system including 6 chillers installed in the building; (<b>c</b>) a power meter for measuring the energy consumption; (<b>d</b>) a networked temperature/humidity sensor module for detecting the necessary parameters for the HVAC control.</p> Full article ">Figure 2
<p>The PIN sensor design for increasing the prediction accuracy based on the conditional probability and Bayes’ theorem.</p> Full article ">Figure 3
<p>The test site simulation and related parameters for building a control model.</p> Full article ">Figure 4
<p>Big data collection for the PIN sensor design and AI control performance test: (<b>a</b>) a sample of one temperature and humidity sensor installed in the 1F lobby; (<b>b</b>) The real time data collection of a day; (<b>c</b>–<b>h</b>) The one-day data can be divided into six segments, and each segment data has different trends and frequencies.</p> Full article ">Figure 4 Cont.
<p>Big data collection for the PIN sensor design and AI control performance test: (<b>a</b>) a sample of one temperature and humidity sensor installed in the 1F lobby; (<b>b</b>) The real time data collection of a day; (<b>c</b>–<b>h</b>) The one-day data can be divided into six segments, and each segment data has different trends and frequencies.</p> Full article ">Figure 5
<p>Two features of big data, including (<b>a</b>) the signal trend, can be expressed by four standard types; (<b>b</b>) one frequency spectrum corresponding to one signal trend.</p> Full article ">Figure 6
<p>Working principles of the AI-assisted HVAC control (<b>a</b>) Trend from the big data collection; (<b>b</b>) Initial guess of the trend by machine learning; (<b>c</b>) Adjustment of the trend filter for the PIN sensor; (<b>d</b>) Frequency output; (<b>e</b>) Comparing with a similar spectrum; (<b>f</b>) Modified trend for the AI-assisted control.</p> Full article ">Figure 7
<p>The energy savings of the AI-assisted control on the hottest week of 2017.</p> Full article ">
15 pages, 6410 KiB  
Article
New Motion Intention Acquisition Method of Lower Limb Rehabilitation Robot Based on Static Torque Sensors
by Yongfei Feng, Hongbo Wang, Luige Vladareanu, Zheming Chen and Di Jin
Sensors 2019, 19(15), 3439; https://doi.org/10.3390/s19153439 - 6 Aug 2019
Cited by 13 | Viewed by 5107
Abstract
The rehabilitation robot is an application of robotic technology for people with limb disabilities. This paper investigates a new applicable and effective sitting/lying lower limb rehabilitation robot (the LLR-Ro). In order to improve the patient’s training initiative and accelerate the rehabilitation process, a [...] Read more.
The rehabilitation robot is an application of robotic technology for people with limb disabilities. This paper investigates a new applicable and effective sitting/lying lower limb rehabilitation robot (the LLR-Ro). In order to improve the patient’s training initiative and accelerate the rehabilitation process, a new motion intention acquisition method based on static torque sensors is proposed. This motion intention acquisition method is established through the dynamics modeling of human–machine coordination, which is built on the basis of Lagrangian equations. Combined with the static torque sensors installed on the mechanism leg joint axis, the LLR-Ro can obtain the active force from the patient’s leg. Based on the variation of the patient’s active force and the kinematic functional relationship of the patient’s leg end point, the patient motion intention is obtained and used in the proposed active rehabilitation training method. The simulation experiment demonstrates the correctness of mechanism leg dynamics equations through ADAMS software and MATLAB software. The calibration experiment of the joint torque sensors’ combining limit range filter with an average value filter provides the hardware support for active rehabilitation training. The consecutive variation of the torque sensors from just the mechanism leg weight, as well as both the mechanism leg and the patient leg weights, obtains the feasibility of lower limb motion intention acquisition. Full article
(This article belongs to the Special Issue Advanced Intelligent Control through Versatile Intelligent Portable Platforms)
►▼ Show Figures

Figure 1

Figure 1
<p>The prototype of the LLR-Ro: (<b>a</b>) The prototype of the LLR-Ro; (<b>b</b>) The movable seat separated from and grouped into the LLR-Ro; (<b>c</b>) The back angle of the movable seat altered from 110° to 170°.</p> Full article ">Figure 2
<p>Design of the mechanism leg.</p> Full article ">Figure 3
<p>The profile and detailed parameters of the torque sensor.</p> Full article ">Figure 4
<p>Design of the hardware control system.</p> Full article ">Figure 5
<p>The motion intention acquisition flow diagram of the patient’s lower limb.</p> Full article ">Figure 6
<p>The linkage model of the mechanism leg.</p> Full article ">Figure 7
<p>Simulation model developed through ADAMS.</p> Full article ">Figure 8
<p>The simulation curves and theoretical curves of the mechanism leg joints: (<b>a</b>) The simulation curves and theoretical curves of the hip joint; (<b>b</b>) The simulation curves and theoretical curves of the knee joint.</p> Full article ">Figure 9
<p>The calibration experiment of torque sensors: (<b>a</b>) The thigh of the mechanism leg is set at the horizontal position; (<b>b</b>) One point is marked from the hip joint axis 585 mm; (<b>c</b>) The analytical weights (each weight is 2.5 kg) are put on the marked point one by one until the weight equals 17.5 kg; (<b>d</b>) The weights start to be unloaded one by one until it equals 0 kg. The steps above are repeated three times, and the voltage values are processed through the combing limit range filter and average value filter.</p> Full article ">Figure 10
<p>The calibration curves of the joint torque sensors: (<b>a</b>) The calibration curves of the hip joint torque sensors; (<b>b</b>) The calibration curves of the knee joint torque sensors.</p> Full article ">Figure 11
<p>The experiment to obtain the joint torques from the mechanism leg weight.</p> Full article ">Figure 12
<p>The calibration curves of the joint torque sensors: (<b>a</b>) Hip torque just from the mechanism leg weight; (<b>b</b>) Knee torque just from the mechanism leg weight.</p> Full article ">Figure 13
<p>The verification experiment of active training without patient active force.</p> Full article ">Figure 14
<p>Joint torques from the mechanism leg and patient leg weights: (<b>a</b>) Hip torque from the mechanism leg and patient leg weights; (<b>b</b>) Knee torque from the mechanism leg and patient leg weights.</p> Full article ">
19 pages, 3081 KiB  
Article
PARSUC: A Parallel Subsampling-Based Method for Clustering Remote Sensing Big Data
by Huiyu Xia, Wei Huang, Ning Li, Jianzhong Zhou and Dongying Zhang
Sensors 2019, 19(15), 3438; https://doi.org/10.3390/s19153438 - 5 Aug 2019
Cited by 10 | Viewed by 3861
Abstract
Remote sensing big data (RSBD) is generally characterized by huge volumes, diversity, and high dimensionality. Mining hidden information from RSBD for different applications imposes significant computational challenges. Clustering is an important data mining technique widely used in processing and analyzing remote sensing imagery. [...] Read more.
Remote sensing big data (RSBD) is generally characterized by huge volumes, diversity, and high dimensionality. Mining hidden information from RSBD for different applications imposes significant computational challenges. Clustering is an important data mining technique widely used in processing and analyzing remote sensing imagery. However, conventional clustering algorithms are designed for relatively small datasets. When applied to problems with RSBD, they are, in general, too slow or inefficient for practical use. In this paper, we proposed a parallel subsampling-based clustering (PARSUC) method for improving the performance of RSBD clustering in terms of both efficiency and accuracy. PARSUC leverages a novel subsampling-based data partitioning (SubDP) method to realize three-step parallel clustering, effectively solving the notable performance bottleneck of the existing parallel clustering algorithms; that is, they must cope with numerous repeated calculations to get a reasonable result. Furthermore, we propose a centroid filtering algorithm (CFA) to eliminate subsampling errors and to guarantee the accuracy of the clustering results. PARSUC was implemented on a Hadoop platform by using the MapReduce parallel model. Experiments conducted on massive remote sensing imageries with different sizes showed that PARSUC (1) provided much better accuracy than conventional remote sensing clustering algorithms in handling larger image data; (2) achieved notable scalability with increased computing nodes added; and (3) spent much less time than the existing parallel clustering algorithm in handling RSBD. Full article
(This article belongs to the Special Issue Computational Intelligence in Remote Sensing)
►▼ Show Figures

Figure 1

Figure 1
<p>Framework of parallel subsampling-based clustering (PARSUC).</p> Full article ">Figure 2
<p>Subsampling-based data partitioning (SubDP) method.</p> Full article ">Figure 3
<p>Pseudocode of the centroid filtering algorithm (CFA).</p> Full article ">Figure 4
<p>The minimum spanning tree (MST) (<b>a</b>) converting to a forest (<b>b</b>).</p> Full article ">Figure 5
<p>The mapping operation of PARSUC.</p> Full article ">Figure 6
<p>The MapReduce-based implementation of PARSUC.</p> Full article ">Figure 7
<p>Variation of sum of squared errors (SSEs) for different combinations of <span class="html-italic">ρ</span> and <span class="html-italic">B</span>.</p> Full article ">Figure 8
<p>Original GaoFen-2 imagery (<b>a</b>), classification results of PARSUC(KM) (<b>b</b>) and K-means (<b>c</b>).</p> Full article ">Figure 9
<p>Ground truth samples, classification results of PARSUC(KM) and K-means at region 1.</p> Full article ">Figure 10
<p>Ground truth samples, classification results of PARSUC(KM) and K-means at region 2.</p> Full article ">Figure 11
<p>Ground truth, water body extraction results of ISODATA and PARSUC(ISO).</p> Full article ">Figure 12
<p>Processing time of PARSUC with different nodes and image sizes.</p> Full article ">Figure 13
<p>Speedup of PARSUC with different number of nodes.</p> Full article ">
15 pages, 4959 KiB  
Article
Monitoring of Chemical Risk Factors for Sudden Infant Death Syndrome (SIDS) by Hydroxyapatite-Graphene-MWCNT Composite-Based Sensors
by Narayanan Sudhan, Nehru Lavanya, Salvatore Gianluca Leonardi, Giovanni Neri and Chinnathambi Sekar
Sensors 2019, 19(15), 3437; https://doi.org/10.3390/s19153437 - 5 Aug 2019
Cited by 9 | Viewed by 4953
Abstract
Sensing properties of chemical sensors based on ternary hydroxyapatite-graphene-multiwalled carbon nanotube (HA-GN-MWCNT) nanocomposite in the detection of chemical substances representing risk factors for sudden infant death syndrome (SIDS), have been evaluated. Characterization data of the synthesized composite have shown that the graphene-MWCNT network [...] Read more.
Sensing properties of chemical sensors based on ternary hydroxyapatite-graphene-multiwalled carbon nanotube (HA-GN-MWCNT) nanocomposite in the detection of chemical substances representing risk factors for sudden infant death syndrome (SIDS), have been evaluated. Characterization data of the synthesized composite have shown that the graphene-MWCNT network serves as a matrix to uniformly disperse the hydroxyapatite nanoparticles and provide suitable electrical properties required for developing novel electrochemical and conductometric sensors. A HA-GN-MWCNT composite-modified glassy carbon electrode (HA-GN-MWCNT/GCE) has been fabricated and tested for the simultaneous monitoring of nicotine and caffeine by cyclic voltammetry (CV) and square wave voltammetry (SWV), whereas a HA-GN-MWCNT conductive gas sensor has been tested for the detection of CO2 in ambient air. Reported results suggest that the synergic combination of the chemical properties of HA and electrical/electrochemical characteristics of the mixed graphene-MWCNT network play a prominent role in enhancing the electrochemical and gas sensing behavior of the ternary HA-GN-MWCNT hybrid nanostructure. The high performances of the developed sensors make them suitable for monitoring unhealthy actions (e.g., smoking, drinking coffee) in breastfeeding women and environmental factors (bad air quality), which are associated with an enhanced risk for SIDS. Full article
(This article belongs to the Special Issue Sensors for Human Safety Monitoring)
►▼ Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Representation of the procedure adopted for the synthesis of HA-GN-MWCNT composite.</p> Full article ">Figure 2
<p>Picture of: (<b>a</b>) electrochemical and (<b>b</b>) conductometric sensor platforms. HA-GN-MWCNT composite sensing film area are indicated by arrows.</p> Full article ">Figure 3
<p>SEM images of (<b>a</b>) pure HA and (<b>b</b> and <b>c</b>) HA-GN-MWCNT. (<b>d</b>) EDX analysis of the HA-GN-MWCNT composite sample.</p> Full article ">Figure 4
<p>XRD patterns of HA-GN-MWCNT composite sample and single components.</p> Full article ">Figure 5
<p>FT-IR of HA-GN-MWCNT composite sample and single components.</p> Full article ">Figure 6
<p>(<b>A</b>) CVs of (<b>A</b>) bare GCE, (b) GO, (c) HA, (d) MWCNT, (e) HA-GN-MWCNT modified GCEs in 1 mM [Fe(CN)<sub>6</sub>]<sup>3−/4−</sup> at a scan rate of 50 mV/s. (<b>B</b>) Electrochemical impedance spectra of the corresponding electrodes recorded at the DC potential of 200 mV, AC potential ±5 mV.</p> Full article ">Figure 7
<p>CVs obtained for the mixture of 1mM of nicotine and caffeine in 0.1 M PBS (pH 7.0) at modified GCEs at a scan rate of 50 mV s<sup>−1</sup>.</p> Full article ">Figure 8
<p>SWVs obtained for the various concentrations of (<b>A</b>) nicotine (0.3-179.5μM) in the presence of 75μM of caffeine and (<b>B</b>) caffeine (4-205μM) in the presence of 20μM of nicotine in 0.1 M PBS (pH 7.0) at HA-GN-MWCNT/GCE. In (<b>C</b>) and (<b>D</b>) are shown the corresponding calibration curves.</p> Full article ">Figure 9
<p>(<b>A</b>) SWVs obtained for various concentrations of nicotine and caffeine (2.35 to 169.35 μM) at HA-GN-MWCNT modified GCE in 0.1 M PBS (pH 7.0). (<b>B</b>) calibration curves.</p> Full article ">Figure 10
<p>SWVs for determination of nicotine and caffeine in milk samples with: (<b>A</b>) without spiking of nicotine and caffeine at HA-MWCNT-GN/GCE in 0.1MPBS at pH 7.0. (<b>B</b>) Standard addition of known concentration of nicotine from 35 µM to 660 µM at HA-MWCNT-GN/GCE in 0.1MPBS at pH 7.0. (<b>C</b>) Standard addition of known concentration of caffeine from 15 µM to 950 µM at HA-MWCNT-GN/GCE in 0.1MPBS at pH 7.0.</p> Full article ">Figure 11
<p>Transient response of the sensor to pulses of CO<sub>2</sub> at the working temperature of (<b>A</b>) 125 °C and (<b>B</b>) 150 °C.</p> Full article ">Figure 12
<p>(<b>A</b>) Transient response of the sensor at the working temperature of 150 °C exposed to CO<sub>2</sub> pulses of different concentration, ranging from 0.25 to 5%; (<b>B</b>) Calibration curve of the investigated sensor in a wide range of concentration (0.015 to 5%).</p> Full article ">
16 pages, 1883 KiB  
Article
Analysis of Byzantine Attacks for Target Tracking in Wireless Sensor Networks
by Yukun Yang, Pengwen Xiong, Qing Wang and Qiang Zhang
Sensors 2019, 19(15), 3436; https://doi.org/10.3390/s19153436 - 5 Aug 2019
Cited by 3 | Viewed by 2908
Abstract
Herein, the problem of target tracking in wireless sensor networks (WSNs) is investigated in the presence of Byzantine attacks. More specifically, we analyze the impact of Byzantine attacks on the performance of a tracking system. First, under the condition of jointly estimating the [...] Read more.
Herein, the problem of target tracking in wireless sensor networks (WSNs) is investigated in the presence of Byzantine attacks. More specifically, we analyze the impact of Byzantine attacks on the performance of a tracking system. First, under the condition of jointly estimating the target state and the attack parameters, the posterior Cramer–Rao lower bound (PCRLB) is calculated. Then, from the perspective of attackers, we define the optimal Byzantine attack and theoretically find a way to achieve such an attack with minimal cost. When the attacked nodes are correctly identified by the fusion center (FC), we further define the suboptimal Byzantine attack and also find a way to realize such an attack. Finally, in order to alleviate the negative impact of attackers on the system performance, a modified sampling importance resampling (SIR) filter is proposed. Simulation results show that the tracking results of the modified SIR filter can be close to the true trajectory of the moving target. In addition, when the quantization level increases, both the security performance and the estimation performance of the tracking system are improved. Full article
(This article belongs to the Section Sensor Networks)
►▼ Show Figures

Figure 1

Figure 1
<p>Simplified system model. The value <math display="inline"><semantics> <mrow> <msub> <mi>η</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> </mrow> </semantics></math> is measurement noise of the <span class="html-italic">i</span>th sensor, <math display="inline"><semantics> <mrow> <msub> <mi>s</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> </mrow> </semantics></math> is the raw measurement, <math display="inline"><semantics> <mrow> <msub> <mi>u</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> </mrow> </semantics></math> is the quantized sensor measurement, and <math display="inline"><semantics> <mrow> <msub> <mi>v</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> </mrow> </msub> </mrow> </semantics></math> is the measurement received by the fusion center, where <span class="html-italic">i</span> = 1, …, <span class="html-italic">N</span>.</p> Full article ">Figure 2
<p>The relationship between <math display="inline"><semantics> <mrow> <msub> <mi>ρ</mi> <mrow> <mi>min</mi> </mrow> </msub> </mrow> </semantics></math> and <span class="html-italic">L</span>.</p> Full article ">Figure 3
<p>Frequency of <math display="inline"><semantics> <mrow> <msub> <mi>v</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mo>=</mo> <mn>0</mn> </mrow> </semantics></math> for all nodes and all time units under optimal Byzantine attacks.</p> Full article ">Figure 4
<p>Frequency of <math display="inline"><semantics> <mrow> <msub> <mi>v</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>t</mi> <mo>+</mo> <mn>1</mn> </mrow> </msub> <mo>=</mo> <mn>0</mn> </mrow> </semantics></math> for all Byzantine nodes and all time units under suboptimal Byzantine attacks.</p> Full article ">Figure 5
<p>Estimation results of the modified SIR filter: (<b>a</b>) The estimated tracks of the moving target when <span class="html-italic">L</span> = 2 and <span class="html-italic">L</span> = 8; and (<b>b</b>) the logarithm of the localization errors.</p> Full article ">Figure 6
<p>The detection rate and the false detection rate of the modified SIR filter: (<b>a</b>) The number of Byzantine nodes that are correctly identified; (<b>b</b>) the number of honest nodes that are misjudged as Byzantines.</p> Full article ">
13 pages, 8463 KiB  
Article
UHF Partial Discharge Location in Power Transformers via Solution of the Maxwell Equations in a Computational Environment
by Luiz A. M. Nobrega, Edson G. Costa, Alexandre J. R. Serres, George V. R. Xavier and Marcus V. D. Aquino
Sensors 2019, 19(15), 3435; https://doi.org/10.3390/s19153435 - 5 Aug 2019
Cited by 19 | Viewed by 4774
Abstract
This paper presents an algorithm for the localisation of partial discharge (PD) sources in power transformers based on the electromagnetic waves radiated by a PD pulse. The proposed algorithm is more accurate than existing methods, since it considers the effects of the reflection, [...] Read more.
This paper presents an algorithm for the localisation of partial discharge (PD) sources in power transformers based on the electromagnetic waves radiated by a PD pulse. The proposed algorithm is more accurate than existing methods, since it considers the effects of the reflection, refractions and diffractions undergone by the ultra-high frequency (UHF) signal within the equipment tank. The proposed method uses computational simulations of the electromagnetic waves generated by PD, and obtains the time delay of the signal between each point in the 3D space and the UHF sensors. The calculated signals can be compared with the signals measured in the field, so that the position of the PD source can be located based on the best correlation between the simulated propagation delay and the measured data. The equations used in the proposed method are defined as a 3D optimisation problem, so that the binary particle swarm optimisation algorithm can be used. To test and demonstrate the proposed algorithm, computational simulations were performed. The solutions were sufficient to identify not only the occurrence of defects, but also the winding and the region (top, centre or base) in which the defect occurred. In all cases, an accuracy of greater than 15 cm was obtained for the location, in a 180 MVA three-phase transformer. Full article
(This article belongs to the Special Issue Acoustic, UHF and RF Sensor Technology for Partial Discharge Detection)
►▼ Show Figures

Figure 1

Figure 1
<p>Positions of the ultra-high frequency (UHF) sensors.</p> Full article ">Figure 2
<p>Representation of the 3D matrix of coordinates.</p> Full article ">Figure 3
<p>Flowchart of the proposed methodology to obtain the transformer propagation times.</p> Full article ">Figure 4
<p>The Gaussian PD current wave.</p> Full article ">Figure 5
<p>Position of simulated defects: (<b>a</b>) Top view of the simulated transformer, and (<b>b</b>) front view of the simulated transformer.</p> Full article ">Figure 6
<p>Model used to test the localisation algorithm.</p> Full article ">Figure 7
<p>Signal propagation time to the sensor located on the left side of the equipment tank (sensor 1): (<b>a</b>) Longitudinal section along the central axis of the transformer, and (<b>b</b>) horizontal section along the central axis of the transformer.</p> Full article ">Figure 8
<p>Magnitude of the signals obtained from the UHF sensors for the defect 6.</p> Full article ">Figure 9
<p>Results of the localisation algorithm: (<b>a</b>) Defect in position 1, (<b>b</b>) defect in position 2, (<b>c</b>) defect in position 3, (<b>d</b>) defect in position 4, (<b>e</b>) defect in position 5, and (<b>f</b>) defect in position 6.</p> Full article ">
20 pages, 2794 KiB  
Article
On-Device Deep Learning Inference for Efficient Activity Data Collection
by Nattaya Mairittha, Tittaya Mairittha and Sozo Inoue
Sensors 2019, 19(15), 3434; https://doi.org/10.3390/s19153434 - 5 Aug 2019
Cited by 11 | Viewed by 5003
Abstract
Labeling activity data is a central part of the design and evaluation of human activity recognition systems. The performance of the systems greatly depends on the quantity and “quality” of annotations; therefore, it is inevitable to rely on users and to keep them [...] Read more.
Labeling activity data is a central part of the design and evaluation of human activity recognition systems. The performance of the systems greatly depends on the quantity and “quality” of annotations; therefore, it is inevitable to rely on users and to keep them motivated to provide activity labels. While mobile and embedded devices are increasingly using deep learning models to infer user context, we propose to exploit on-device deep learning inference using a long short-term memory (LSTM)-based method to alleviate the labeling effort and ground truth data collection in activity recognition systems using smartphone sensors. The novel idea behind this is that estimated activities are used as feedback for motivating users to collect accurate activity labels. To enable us to perform evaluations, we conduct the experiments with two conditional methods. We compare the proposed method showing estimated activities using on-device deep learning inference with the traditional method showing sentences without estimated activities through smartphone notifications. By evaluating with the dataset gathered, the results show our proposed method has improvements in both data quality (i.e., the performance of a classification model) and data quantity (i.e., the number of data collected) that reflect our method could improve activity data collection, which can enhance human activity recognition systems. We discuss the results, limitations, challenges, and implications for on-device deep learning inference that support activity data collection. Also, we publish the preliminary dataset collected to the research community for activity recognition. Full article
(This article belongs to the Special Issue Mobile Sensing: Platforms, Technologies and Challenges)
►▼ Show Figures

Figure 1

Figure 1
<p>The system architecture of the proposed on-device deep learning inference for efficient activity data collection works as following (<b>a</b>) we first train activity labels with mobile sensors using an long short-term memory (LSTM) for recognition model and deploy it for on-device inference, (<b>b</b>) we collect mobile sensors and activity labels on a smartphone from a user, (<b>c</b>) the smartphone detects an activity that the user is doing by using an on-device deep learning inference model adopted, (<b>d</b>) the user receives information about the estimated activity as feedback (e.g., the notification is showing that “are you walking?” means the device is on a user who is walking), (<b>e</b>) the user repeats the process of activity data collection efficiently, and (<b>f</b>) we finally obtain accurate activity dataset.</p> Full article ">Figure 2
<p>A schematic diagram of the proposed LSTM-based deep learning model for activity recognition system works as following (<b>a</b>) the inputs are raw signals obtained from acceleration sensors, (<b>b</b>) segment into windows of length <span class="html-italic">T</span>, (<b>c</b>) fed into LSTM-based deep learning model, (<b>d</b>) and finally, the model outputs class prediction for each time step.</p> Full article ">Figure 3
<p>Many-to-one long short-term memory (LSTM) network architecture used for activity classification with six classes. n stands for the number of samples included in a 2.56 s window.</p> Full article ">Figure 4
<p>Training session’s progress over iterations.</p> Full article ">Figure 5
<p>The results for a classifier of the LSTM model.</p> Full article ">Figure 6
<p>FahLog: a mobile app for collecting sensor data and activity labels.</p> Full article ">Figure 7
<p>Steps to show a estimated activity as feedback to a user.</p> Full article ">Figure 8
<p>The number of activity labels for each method.</p> Full article ">Figure 9
<p>The average classification performance of all models for each method.</p> Full article ">Figure 10
<p>The F1-score performance results of several machine learning models.</p> Full article ">Figure 11
<p>The precision performance results of several machine learning models.</p> Full article ">Figure 12
<p>The recall performance results of several machine learning models.</p> Full article ">
14 pages, 2611 KiB  
Article
Congestion Control in CoAP Observe Group Communication
by Chanwit Suwannapong and Chatchai Khunboa
Sensors 2019, 19(15), 3433; https://doi.org/10.3390/s19153433 - 5 Aug 2019
Cited by 26 | Viewed by 4708 | Correction
Abstract
The Constrained Application Protocol (CoAP) is a simple and lightweight machine-to-machine (M2M) protocol for constrained devices for use in lossy networks which offers a small memory capacity and limited processing. Designed and developed by the Internet Engineering Task Force (IETF), it functions as [...] Read more.
The Constrained Application Protocol (CoAP) is a simple and lightweight machine-to-machine (M2M) protocol for constrained devices for use in lossy networks which offers a small memory capacity and limited processing. Designed and developed by the Internet Engineering Task Force (IETF), it functions as an application layer protocol and benefits from reliable delivery and simple congestion control. It is implemented for request/response message exchanges over the User Datagram Protocol (UDP) to support the Internet of Things (IoT). CoAP also provides a basic congestion control mechanism. In dealing with its own congestion, it relies on a fixed interval retransmission timeout (RTO) and binary exponential backoff (BEB). However, the default CoAP congestion control is considered to be unable to effectively perform group communication and observe resources, and it cannot handle rapid, frequent requests. This results in buffer overflow and packet loss. To overcome these problems, we proposed a new congestion control mechanism for CoAP Observe Group Communication, namely Congestion Control Random Early Detection (CoCo-RED), consisting of (1) determining and calculating an RTO timer, (2) a Revised Random Early Detection (RevRED) algorithm which has recently been developed and primarily based on the buffer management of TCP congestion control, and (3) a Fibonacci Pre-Increment Backoff (FPB) algorithm which waits for backoff time prior to retransmission. All the aforementioned algorithms were therefore implemented instead of the default CoAP mechanism. In this study, evaluations were carried out regarding the efficiency of the developed CoCo-RED using a Cooja simulator. The congestion control mechanism can quickly handle the changing behaviors of network communication, and thus it prevents the buffer overflow that leads to congestions. The results of our experiments indicate that CoCo-RED can control congestion more effectively than the default CoAP in every condition. Full article
(This article belongs to the Section Sensor Networks)
►▼ Show Figures

Figure 1

Figure 1
<p>(<b>a</b>) Constrained Application Protocol (CoAP)-based message transmission; (<b>b</b>) Observing Resource message transmission. CON: confirmable; ACK: acknowledgement.</p> Full article ">Figure 2
<p>(<b>a</b>) Default CoAP-CON mode; (<b>b</b>) Observing Resource-CON mode. RTO: retransmission timeout; BEB: binary exponential backoff.</p> Full article ">Figure 3
<p>Congestion Control Random Early Detection (CoCo-RED) thresholds in the buffer queue.</p> Full article ">Figure 4
<p>CoCo-RED timing diagram. FPB: Fibonacci Pre-Increment Backoff.</p> Full article ">Figure 5
<p>Backoff method durations for one message transmission and four message retransmissions, starting with <span class="html-italic">RTO<sub>init</sub></span> = 2 s.</p> Full article ">Figure 6
<p>Packet drop probability function for CoCo-RED.</p> Full article ">Figure 7
<p>An overview of the RTO used to maintain and update the RTO state information for a destination endpoint in CoCo-RED.</p> Full article ">Figure 8
<p>The five network topologies used for performance analysis (chain, grid, cross, dumbbell and random). The distance between the neighboring nodes is 10 m. The gray nodes are the RPL border routers. The orange nodes are the clients and the gold nodes are the groups of servers for the Obs messages.</p> Full article ">
14 pages, 1768 KiB  
Article
A Unique Interactive Nanostructure Knitting based Passive Sampler Adsorbent for Monitoring of Hg2+ in Water
by Raghuraj S. Chouhan, Gregor Žitko, Vesna Fajon, Igor Živković, Majda Pavlin, Sabina Berisha, Ivan Jerman, Alenka Vesel and Milena Horvat
Sensors 2019, 19(15), 3432; https://doi.org/10.3390/s19153432 - 5 Aug 2019
Cited by 9 | Viewed by 4146
Abstract
This work reports the development of ultralight interwoven ultrathin graphitic carbon nitride (g-CN) nanosheets for use as a potential adsorbent in a passive sampler (PAS) designed to bind Hg2+ ions. The g-CN nanosheets were prepared from bulk g-CN synthesised via a modified [...] Read more.
This work reports the development of ultralight interwoven ultrathin graphitic carbon nitride (g-CN) nanosheets for use as a potential adsorbent in a passive sampler (PAS) designed to bind Hg2+ ions. The g-CN nanosheets were prepared from bulk g-CN synthesised via a modified high-temperature short-time (HTST) polycondensation process. The crystal structure, surface functional groups, and morphology of the g-CN nanosheets were characterised using a battery of instruments. The results confirmed that the as-synthesized product is composed of few-layered nanosheets. The adsorption efficiency of g-CN for binding Hg2+ (100 ng mL−1) in sea, river, rain, and Milli-Q quality water was 89%, 93%, 97%, and 100%, respectively, at natural pH. Interference studies found that the cations tested (Co2+, Ca2+, Zn2+, Fe2+, Mn2+, Ni2+, Bi3+, Na+, and K+) had no significant effect on the adsorption efficiency of Hg2+. Different parameters were optimised to improve the performance of g-CN such as pH, contact time, and amount of adsorbent. Optimum conditions were pH 7, 120 min incubation time and 10 mg of nanosheets. The yield of nanosheets was 72.5%, which is higher compared to other polycondensation processes using different monomers. The g-CN sheets could also be regenerated up to eight times with only a 20% loss in binding efficiency. Overall, nano-knitted g-CN is a promising low-cost green adsorbent for use in passive samplers or as a transducing material in sensor applications. Full article
(This article belongs to the Special Issue Global Mercury Assessment Sensing Strategies)
►▼ Show Figures

Figure 1

Figure 1
<p>Morphological characteristics of as-synthesized graphitic carbon nitride (g-CN) nanosheets. SEM images of g-CN nanosheets showing few-layered lamellar piled together (<b>a</b> and <b>b</b>). HR-TEM image of g-CN with low (<b>c</b>) and high (<b>d</b>) magnification. AFM images of g-CN nanosheets and the corresponding height profiles of different regions (<b>e</b> and <b>f</b>).</p> Full article ">Figure 2
<p>Structural characteristics of as-synthesized nanosheets. (<b>a</b>) FTIR spectra of g-CN nanosheets at the frequency range of 500–4000 cm<sup>−1</sup>, (<b>b</b>) XRD spectra of g-CN nanosheets, (<b>c</b>) Raman signature profile of the nanosheets with distinct D and G bands, (<b>d</b>) XPS high resolution scan of C1, (<b>e</b>) N1 and (<b>f</b>) full survey spectrum shows three peaks of carbon, nitrogen and oxygen.</p> Full article ">Figure 3
<p>The % recovery vs pH (2–10) under identical conditions.</p> Full article ">Figure 4
<p>Effect of contact time on the binding of Hg<sup>2+</sup> (100 ng mL<sup>−1</sup>) under ideal conditions. Agitation speed = 210 (rpm), room temperature = 21 °C and optimized pH-7.</p> Full article ">Figure 5
<p>Optimal studies of g-CN nanosheets at (<b>a</b>) different concentration of g-CN (5, 10, 20, and 40 mg mL<sup>−1</sup>) required to saturate Hg<sup>2+</sup> (100 ng mL<sup>−1</sup>) under optimised incubation and (<b>b</b>) binding efficiency of g-CN nanosheets (10 mg mL<sup>−1</sup>) at different concentration of Hg<sup>2+</sup> (pH 7). The sample volume was 3 mL.</p> Full article ">Figure 6
<p>Adsorption studies using sea, river, rain, and Milli-Q water spiked with 100 ng mL<sup>−1</sup> of Hg<sup>2+</sup>. The sample volume was 3 mL.</p> Full article ">Figure 7
<p>Potential influence of metal ions on the % recovery of Hg<sup>2+</sup> (100 ng mL<sup>−1</sup> Hg<sup>2+</sup> and 500 ng mL<sup>−1</sup> of Co<sup>2+</sup>, Ca<sup>2+</sup>, Zn<sup>2+</sup>, Fe<sup>2+</sup>, Mn<sup>2+</sup>, Ni<sup>2+</sup>, Bi<sup>3+</sup>, Na<sup>+</sup>, and K<sup>+</sup>).</p> Full article ">Figure 8
<p>Adsorption–desorption regeneration cycling of g-CN.</p> Full article ">Scheme 1
<p>Schematic representation of the present work. (<b>a</b>) synthesis of nanosheets using the high-temperature short-time (HTST) polycondensation process, (<b>b</b>) overall method used to analyze Hg<sup>2+</sup>.</p> Full article ">
16 pages, 6244 KiB  
Article
A Hybrid Two-Axis Force Sensor for the Mesoscopic Structural Superlubricity Studies
by Taotao Sun, Zhanghui Wu, Zhihong Li, Quanshui Zheng and Li Lin
Sensors 2019, 19(15), 3431; https://doi.org/10.3390/s19153431 - 5 Aug 2019
Cited by 2 | Viewed by 4214
Abstract
Structural superlubricity (SSL) is a state of nearly zero friction and zero wear between two directly contacted solid surfaces. Recently, SSL was achieved in mesoscale and thus opened the SSL technology which promises great applications in Micro-electromechanical Systems (MEMS), sensors, storage technologies, etc. [...] Read more.
Structural superlubricity (SSL) is a state of nearly zero friction and zero wear between two directly contacted solid surfaces. Recently, SSL was achieved in mesoscale and thus opened the SSL technology which promises great applications in Micro-electromechanical Systems (MEMS), sensors, storage technologies, etc. However, load issues in current mesoscale SSL studies are still not clear. The great challenge is to simultaneously measure both the ultralow shear forces and the much larger normal forces, although the widely used frictional force microscopes (FFM) and micro tribometers can satisfy the shear forces and normal forces requirements, respectively. Here we propose a hybrid two-axis force sensor that can well fill the blank between the capabilities of FFM and micro tribometers for the mesoscopic SSL studies. The proposed sensor can afford 1mN normal load with 10 nN lateral resolution. Moreover, the probe of the sensor is designed at the edge of the structure for the convenience of real-time optical observation. Calibrations and preliminary experiments are conducted to validate the performance of the design. Full article
(This article belongs to the Special Issue Sensors in Experimental Mechanics)
►▼ Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Summary of the current commercial instruments for tribology studies. The shade shows the requirements of forces in SSL studies. Force ranges of the instruments are enclosed in the rounded squares. There is a blank of the general tools which is circled by the dotted line.</p> Full article ">Figure 2
<p>Schematic model of the hybrid two-axis force sensor.</p> Full article ">Figure 3
<p>Schematic of sensing structures. (<b>a</b>,<b>b</b>) Structure for the normal load detection. (<b>a</b>) A 3D model of the double-(leaf-)cantilever pair; (<b>b</b>) a side view sketch of the differential capacitor configuration. (<b>c</b>,<b>d</b>) Structure for the lateral load detection. (<b>c</b>) A 3D model of the teeter-totter structure; (<b>d</b>) a side view sketch of the differential capacitor configuration.</p> Full article ">Figure 4
<p>FEA on teeter-totter structure. (<b>a</b>) Loads applied on the structure—100 μN lateral force and 1 mN normal load; (<b>b</b>) meshing result, locally refined; (<b>c</b>) simulation result of the max equivalent stress.</p> Full article ">Figure 5
<p>Geometric illustration of the coupling from the teeter-totter structure. (<b>a</b>) A side view model of the teeter-totter structure with a tip connected on the plate by glue; (<b>b</b>) A simplified model from (<b>a</b>); (<b>c</b>) Two common situations (forward and backward) during the friction measurements.</p> Full article ">Figure 6
<p>Photos of the sensor and the test system. (<b>a</b>) A photo of the hybrid two-axis force sensor; (<b>b</b>) A SEM picture of the micro teeter-totter sensing part; (<b>c</b>) A photo of the test system. 1-the hybrid force sensor, 2-vertical moving stages, 3-optical microscope (Hirox), 4-horizontal moving stages, 5-a test sample. The inset picture at the upper right corner is a close look at the microscope, the sample, the lateral force sensor and the adapter boards.</p> Full article ">Figure 7
<p>Calibration results. (<b>a</b>,<b>b</b>) Relationships between the digital capacitance signals and the reference forces. The inset picture in each picture shows the peak-to-peak noise. The red line is the curve fitting result. (<b>a</b>) The normal load sensing part; (<b>b</b>) the lateral force sensing part. The grey dotted line represents a force value of 25 μN, which is the recommended range for a better linearity. (<b>c</b>) Geometric coupling calibration. Filled icons are the theoretical results; hollow icons are the experimental results. The red line is the curve fitting. The diagram illustrates the relationship between the applied normal load and the induced lateral signal. (<b>d</b>) Assembly error tests. The black columns show the experiments without wires. The grey-grade columns show the experiments with wires. Each height of the column represents an average calibrated force constant (<math display="inline"><semantics> <mrow> <mo>△</mo> <mi>N</mi> <mo>/</mo> <mo>△</mo> <msub> <mi>C</mi> <mi>N</mi> </msub> </mrow> </semantics></math>) of each assembly. Error bars are shown to represent the std within each assembly.</p> Full article ">Figure 8
<p>Friction experiments. (<b>a</b>) Photos observed by the optical microscope. The shoe-like shape in a dark purple color is the graphene flake; the blue square pointed out by an arrow is the 5 μm × 5 μm graphite mesa, the tip is shown in a bright triangle shape at the bottom of each photo. A small sliding distance of the graphite mesa can be observed after applying a load on it by the tip; (<b>b</b>) friction loops of the graphite-mica interface using the #3 teeter-totter beam; (<b>c</b>) friction loops of the graphite-graphene interface using the #5 teeter-totter beam. (<b>b</b>,<b>c</b>) The picture at the bottom shows a magnified friction loop under the smallest normal load in each case. Points labeled in black squares and white triangles represent signals of different moving directions. The red thick line is the average lateral force in each direction; (<b>d</b>) relationships between the normal forces and the friction forces. The insert diagram is a magnification of the graphite-graphene case.</p> Full article ">
23 pages, 1239 KiB  
Article
Smartphone-Based Platform for Affect Monitoring through Flexibly Managed Experience Sampling Methods
by Carlos Bailon, Miguel Damas, Hector Pomares, Daniel Sanabria, Pandelis Perakakis, Carmen Goicoechea and Oresti Banos
Sensors 2019, 19(15), 3430; https://doi.org/10.3390/s19153430 - 5 Aug 2019
Cited by 10 | Viewed by 5291
Abstract
The identification of daily life events that trigger significant changes on our affective state has become a fundamental task in emotional research. To achieve it, the affective states must be assessed in real-time, along with situational information that could contextualize the affective data [...] Read more.
The identification of daily life events that trigger significant changes on our affective state has become a fundamental task in emotional research. To achieve it, the affective states must be assessed in real-time, along with situational information that could contextualize the affective data acquired. However, the objective monitoring of the affective states and the context is still in an early stage. Mobile technologies can help to achieve this task providing immediate and objective data of the users’ context and facilitating the assessment of their affective states. Previous works have developed mobile apps for monitoring affective states and context, but they use a fixed methodology which does not allow for making changes based on the progress of the study. This work presents a multimodal platform which leverages the potential of the smartphone sensors and the Experience Sampling Methods (ESM) to provide a continuous monitoring of the affective states and the context in an ubiquitous way. The platform integrates several elements aimed to expedite the real-time management of the ESM questionnaires. In order to show the potential of the platform, and evaluate its usability and its suitability for real-time assessment of affective states, a pilot study has been conducted. The results demonstrate an excellent usability level and a good acceptance from the users and the specialists that conducted the study, and lead to some suggestions for improving the data quality of mobile context-aware ESM-based systems. Full article
(This article belongs to the Special Issue Selected Papers from UCAmI 2018 –The 12th International Conference on Ubiquitous Computing and Ambient Intelligence)
►▼ Show Figures

Figure 1

Figure 1
<p>Number of searches of “mood app” during the last 10 years. Values expressed in percentage relative to the total amount of searches on that topic. Source: Google Trends [<a href="#B17-sensors-19-03430" class="html-bibr">17</a>].</p> Full article ">Figure 2
<p>Architecture of the monitoring platform.</p> Full article ">Figure 3
<p>Russell’s circumplex model of mood [<a href="#B14-sensors-19-03430" class="html-bibr">14</a>].</p> Full article ">Figure 4
<p>Screenshots of the ESM questions for assessing valence and arousal.</p> Full article ">Figure 5
<p>ESM Management Interface.</p> Full article ">Figure 6
<p>Percentage of questionnaires answered (blue), expired (orange) and actively dismissed (grey) for each participant during the entire study.</p> Full article ">Figure 7
<p>Overall percentage of questionnaires answered (blue), expired (orange) and actively dismissed (grey) per interval of daily hours.</p> Full article ">Figure 8
<p>Overall response rate registered per day of study. The red dashed vertical line splits the graphic in the two weeks of the study.</p> Full article ">Figure 9
<p>Completion times of the questionnaires per day of study for the valence question. Times over 300 s have not been considered.</p> Full article ">Figure 10
<p>Completion times of the questionnaires per day of study for the arousal question. Times over 300 s have not been considered.</p> Full article ">Figure 11
<p>Time elapsed from the reception of the notification to the participant’s response per interval of daily hours.</p> Full article ">Figure 12
<p>System Usability Scale (SUS) score obtained from each participant. The horizontal lines represent the mean SUS score of the system and the threshold value that indicates a good usability.</p> Full article ">
26 pages, 8003 KiB  
Article
Real-Time Robust and Optimized Control of a 3D Overhead Crane System
by Arash Khatamianfar and Andrey V. Savkin
Sensors 2019, 19(15), 3429; https://doi.org/10.3390/s19153429 - 5 Aug 2019
Cited by 12 | Viewed by 7463
Abstract
A new and advanced control system for three-dimensional (3D) overhead cranes is proposed in this study using state feedback control in discrete time to deliver high performance trajectory tracking with minimum load swings in high-speed motions. By adopting the independent joint control strategy, [...] Read more.
A new and advanced control system for three-dimensional (3D) overhead cranes is proposed in this study using state feedback control in discrete time to deliver high performance trajectory tracking with minimum load swings in high-speed motions. By adopting the independent joint control strategy, a new and simplified model is developed where the overhead crane actuators are used to design the controller, with all the nonlinear equations of motions being viewed as disturbances affecting each actuator. A feedforward control is then designed to tackle these disturbances via computed torque control technique. A new load swing control is designed along with a new motion planning scheme to robustly minimize load swings as well as allowing fast load transportation without violating system’s constraints through updating reference trolley accelerations. The stability and performance analysis of the proposed discrete-time control system are demonstrated and validated analytically and practically. Full article
(This article belongs to the Special Issue Sensors and Robot Control)
►▼ Show Figures

Figure 1

Figure 1
<p>Schematic structure for a 3D overhead crane [<a href="#B53-sensors-19-03429" class="html-bibr">53</a>].</p> Full article ">Figure 2
<p>Electromechanical schematic for an armature-controlled PM DC motor.</p> Full article ">Figure 3
<p>Configuration of the proposed control system for 3D overhead crane [<a href="#B53-sensors-19-03429" class="html-bibr">53</a>].</p> Full article ">Figure 4
<p>LSPB trajectories [<a href="#B49-sensors-19-03429" class="html-bibr">49</a>]: (<b>a</b>) Position; (<b>b</b>) velocity; (<b>c</b>) acceleration profiles for traveling and traversing motions; (<b>d</b>) Position; (<b>e</b>) velocity, and (<b>f</b>) acceleration profiles for hoisting motion [<a href="#B52-sensors-19-03429" class="html-bibr">52</a>].</p> Full article ">Figure 5
<p>Comparison between original and modified reference traveling trajectory; (<b>a</b>) Position profile; (<b>b</b>) Velocity profile; (<b>c</b>) Acceleration profile.</p> Full article ">Figure 6
<p>The proposed motion planning flowchart. (<b>a</b>) Calculation of correction velocities; (<b>b</b>) Calculation of correction acceleration, recalculation of correction velocities, hoisting velocity, and acceleration if decelerating time is extended.</p> Full article ">Figure 7
<p>(<b>a</b>) The experimental overhead crane setup used in this study; (<b>b</b>) Zoomed-in view.</p> Full article ">Figure 8
<p>Actual and reference trajectories: (<b>a</b>) Traveling; (<b>b</b>) traversing, and (<b>c</b>) hoisting for slow trajectory; (<b>d</b>) Traveling; (<b>e</b>) traversing, and (<b>f</b>) hoisting for fast trajectory under Scenario III.</p> Full article ">Figure 9
<p>Control input voltages: (<b>a</b>) Slow trajectory; (<b>b</b>) Fast trajectory.</p> Full article ">Figure 10
<p>(<b>a</b>,<b>b</b>): Estimation error of swing angle and their velocities estimates for slow trajectory, respectively; (<b>c</b>,<b>d</b>): Estimation error of swing angle and their velocities estimates for fast trajectory, respectively.</p> Full article ">Figure 11
<p>Load swing in <span class="html-italic">X</span> and <span class="html-italic">Y</span> directions for slow trajectory in (<b>a</b>) 1st; (<b>b</b>) 2nd; (<b>c</b>) 3rd Scenarios and for fast trajectory in (<b>d</b>) 1st; (<b>e</b>) 2nd, and (<b>f</b>) 3rd Scenarios.</p> Full article ">Figure 12
<p>Original and modified reference trajectories in 3rd Scenario for fast trajectory: (<b>a</b>) Positions; (<b>b</b>) velocities; (<b>c</b>) accelerations for traveling motion; (<b>d</b>) Positions; (<b>e</b>) velocities, and (<b>f</b>) accelerations for traversing motion.</p> Full article ">Figure 12 Cont.
<p>Original and modified reference trajectories in 3rd Scenario for fast trajectory: (<b>a</b>) Positions; (<b>b</b>) velocities; (<b>c</b>) accelerations for traveling motion; (<b>d</b>) Positions; (<b>e</b>) velocities, and (<b>f</b>) accelerations for traversing motion.</p> Full article ">Figure 13
<p>Trajectory tracking error for slow trajectory in (<b>a</b>) 1st; (<b>b</b>) 2nd; (<b>c</b>) 3rd scenarios, and for fast trajectory in (<b>d</b>) 1st, (<b>e</b>) 2nd; and (<b>f</b>) 3rd Scenarios.</p> Full article ">
13 pages, 338 KiB  
Article
A Node Density Control Learning Method for the Internet of Things
by Shumei Lou, Gautam Srivastava and Shuai Liu
Sensors 2019, 19(15), 3428; https://doi.org/10.3390/s19153428 - 5 Aug 2019
Cited by 19 | Viewed by 4047
Abstract
When examining density control learning methods for wireless sensor nodes, control time is often long and power consumption is usually very high. This paper proposes a node density control learning method for wireless sensor nodes and applies it to an environment based on [...] Read more.
When examining density control learning methods for wireless sensor nodes, control time is often long and power consumption is usually very high. This paper proposes a node density control learning method for wireless sensor nodes and applies it to an environment based on Internet of Things architectures. Firstly, the characteristics of wireless sensors networks and the structure of mobile nodes are analyzed. Combined with the flexibility of wireless sensor networks and the degree of freedom of real-time processing and configuration of field programmable gate array (FPGA) data, a one-step transition probability matrix is introduced. In addition, the probability of arrival of signals between any pair of mobile nodes is also studied and calculated. Finally, the probability of signal connection between mobile nodes is close to 1, approximating the minimum node density at T. We simulate using a fully connected network identifying a worst-case test environment. Detailed experimental results show that our novel proposed method has shorter completion time and lower power consumption than previous attempts. We achieve high node density control as well at close to 90%. Full article
(This article belongs to the Special Issue Big Data Driven IoT for Smart Cities)
►▼ Show Figures

Figure 1

Figure 1
<p>Wireless Sensor Node Schematic.</p> Full article ">Figure 2
<p>Comparison of node density control completion time.</p> Full article ">
11 pages, 2980 KiB  
Article
Low-Cost and Highly Sensitive Wearable Sensor Based on Napkin for Health Monitoring
by Liping Xie, Peng Chen, Shuo Chen, Kun Yu and Hongbin Sun
Sensors 2019, 19(15), 3427; https://doi.org/10.3390/s19153427 - 5 Aug 2019
Cited by 29 | Viewed by 5922
Abstract
The development of sensors with high sensitivity, good flexibility, low cost, and capability of detecting multiple inputs is of great significance for wearable electronics. Herein, we report a napkin-based wearable capacitive sensor fabricated by a novel, low-cost, and facile strategy. The capacitive sensor [...] Read more.
The development of sensors with high sensitivity, good flexibility, low cost, and capability of detecting multiple inputs is of great significance for wearable electronics. Herein, we report a napkin-based wearable capacitive sensor fabricated by a novel, low-cost, and facile strategy. The capacitive sensor is composed of two pieces of electrode plates manufactured by spontaneous assembly of silver nanowires (NWs) on a polydimethylsiloxane (PDMS)-patterned napkin. The sensor possesses high sensitivity (>7.492 kPa−1), low cost, and capability for simultaneous detection of multiple signals. We demonstrate that the capacitive sensor can be applied to identify a variety of human physiological signals, including finger motions, eye blinking, and minute wrist pulse. More interestingly, the capacitive sensor comfortably attached to the temple can simultaneously monitor eye blinking and blood pulse. The demonstrated sensor shows great prospects in the applications of human–machine interface, prosthetics, home-based healthcare, and flexible touch panels. Full article
(This article belongs to the Special Issue Wearable Soft Sensors)
►▼ Show Figures

Graphical abstract

Graphical abstract
Full article ">Figure 1
<p>Schematic illustration of the fabrication of the napkin-based capacitive sensor. (<b>a</b>) Fabrication process. Step 1: A folded paper with hollowed pattern coated with polydimethylsiloxane (PDMS). Step 2: A piece of napkin is inserted into the folded paper, resulting in hydrophilic and hydrophobic pattern on the napkin. Step 3: Adding silver nanowires (NWs) in ethanol solution to the patterned napkin. Step 4: Packing it with PU film. Step 5: Two pieces of the electrode plates are coupling to form a capacitive sensor. (<b>b</b>) Photographs of the folded paper, hydrophilic–hydrophobic patterned napkin, single-electrode plate, and capacitive sensor.</p> Full article ">Figure 2
<p>Optimization of the areal density of the silver NWs on a napkin. SEM images of the silver NWs on napkins with the density of 0.67 mg cm<sup>−2</sup> (<b>a</b>), 1 mg cm<sup>−2</sup> (<b>b</b>), and 1.33 mg cm<sup>−2</sup> (<b>c</b>), respectively. The scale bars are 1 μm.</p> Full article ">Figure 3
<p>Optimization and characterization of the electrical properties and performances of the capacitive sensor. (<b>a</b>) Changes in capacitance under pressure on and off. (<b>b</b>) Relative change in capacitance with applied pressure. Error bars show standard deviation with N = 3.</p> Full article ">Figure 4
<p>Capacitive sensor for detection of finger motion and eye blinking. (<b>a</b>) Capacitance profile of finger motion. (<b>b</b>) Capacitance profile of eye blinking. (<b>c</b>) The 5th level wavelet high-frequency coefficient decomposed from the original eye blinking signal (d5).</p> Full article ">Figure 5
<p>Detection and analysis of wrist pulse acquired by the capacitive sensor. (<b>a</b>) The recorded wrist pulse. (<b>b</b>) The wrist pulse after denoising and drift removal. (<b>c</b>) The 5th level wavelet high-frequency coefficient decomposed from the recorded wrist pulse signal (d5). (<b>d</b>) The enlarged denoising wrist pulse.</p> Full article ">
4 pages, 441 KiB  
Opinion
Digital Health Sensing for Personalized Dermatology
by Pier Spinazze, Alex Bottle and Josip Car
Sensors 2019, 19(15), 3426; https://doi.org/10.3390/s19153426 - 5 Aug 2019
Cited by 8 | Viewed by 4858
Abstract
The rapid evolution of technology, sensors and personal digital devices offers an opportunity to acquire health related data seamlessly, unobtrusively and in real time. In this opinion piece, we discuss the relevance and opportunities for using digital sensing in dermatology, taking eczema as [...] Read more.
The rapid evolution of technology, sensors and personal digital devices offers an opportunity to acquire health related data seamlessly, unobtrusively and in real time. In this opinion piece, we discuss the relevance and opportunities for using digital sensing in dermatology, taking eczema as an exemplar. Full article
(This article belongs to the Section Intelligent Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>Example of integrated data collection and evaluation using a smartphone.</p> Full article ">
32 pages, 5591 KiB  
Review
Recent Advances in Electric-Double-Layer Transistors for Bio-Chemical Sensing Applications
by Ning Liu, Ru Chen and Qing Wan
Sensors 2019, 19(15), 3425; https://doi.org/10.3390/s19153425 - 5 Aug 2019
Cited by 50 | Viewed by 16344
Abstract
As promising biochemical sensors, ion-sensitive field-effect transistors (ISFETs) are used widely in the growing field of biochemical sensing applications. Recently, a new type of field-effect transistor gated by ionic electrolytes has attracted intense attention due to the extremely strong electric-double-layer (EDL) gating effect. [...] Read more.
As promising biochemical sensors, ion-sensitive field-effect transistors (ISFETs) are used widely in the growing field of biochemical sensing applications. Recently, a new type of field-effect transistor gated by ionic electrolytes has attracted intense attention due to the extremely strong electric-double-layer (EDL) gating effect. In such devices, the carrier density of the semiconductor channel can be effectively modulated by an ion-induced EDL capacitance at the semiconductor/electrolyte interface. With advantages of large specific capacitance, low operating voltage and sensitive interfacial properties, various EDL-based transistor (EDLT) devices have been developed for ultrasensitive portable sensing applications. In this article, we will review the recent progress of EDLT-based biochemical sensors. Starting with a brief introduction of the concepts of EDL capacitance and EDLT, we describe the material compositions and the working principle of EDLT devices. Moreover, the biochemical sensing performances of several important EDLTs are discussed in detail, including organic-based EDLTs, oxide-based EDLTs, nanomaterial-based EDLTs and neuromorphic EDLTs. Finally, the main challenges and development prospects of EDLT-based biochemical sensors are listed. Full article
(This article belongs to the Special Issue Potentiometric Bio/Chemical Sensing)
►▼ Show Figures

Figure 1

Figure 1
<p>Number of publications and total citations on EDLT-based sensors sorted by year with key phrases “electrolyte-gated field-effect transistor sensor” or “electric-double-layer transistor sensor” searched in the Web of Science on 20 July 2019.</p> Full article ">Figure 2
<p>Stern model of the EDL generated on a positively charged electrode surface. Adapted from Ref. [<a href="#B30-sensors-19-03425" class="html-bibr">30</a>].</p> Full article ">Figure 3
<p>Schematic diagrams of (<b>a</b>) a top gate EDLT and (<b>b</b>) a side gate EDLT.</p> Full article ">Figure 4
<p>(<b>a</b>) The charge process in EDL formation under an external voltage. (<b>b</b>) The formation of EDL capacitors at the electrode/electrolyte interfaces. (<b>c</b>) The discharge process of EDL capacitors. Adapted from [<a href="#B29-sensors-19-03425" class="html-bibr">29</a>].</p> Full article ">Figure 5
<p>(<b>a</b>) Schematic of the solution-gated α6T-based EDLT sensor. (<b>b</b>) Transfer curves of unmodified, UV-oxidized and APTES-modified EDLTs, measured at pH 5. (<b>c</b>) Shifts in Vth of unmodified, UV-oxidized and APTES-modified EDLTs as a function of solution pH values. (<b>d</b>) Calibration of Vth shifts for APTES-modified device and oxidized device by subtracting the response of untreated α6T-based EDLT. (<b>e</b>) The responses of current with lapsed time to the addition of penicillin for the α6T-based EDLT functionalized with APTES/PEN (green line), and the untreated device (grey line). (<b>f</b>) Vth shifts against penicillin concentration for as-prepared devices with physisorbed enzymes, chemisorbed enzymes and after three washing steps. The dashed line indicates the simulated response of Vth shifts to pH variations in an unbuffered electrolyte. Adapted from [<a href="#B86-sensors-19-03425" class="html-bibr">86</a>].</p> Full article ">Figure 6
<p>(<b>a</b>) Schematic of the solution-gated organic EDLT with the charge arrangements in a PL/SA/Ab/CRP multilayer shown on the right. The equivalent circuit of the capacitances involved is also shown. (<b>b</b>) Effect of the ionic strength on the relative current variations for the PL/SA(AV)/Ab and (<b>c</b>) PL/SA(AV)/Ab/CRP multilayers. (<b>d</b>) Effect of the ionic strength on the relative capacitance changes for the PL/SA(AV)/Ab and (<b>e</b>) PL/SA(AV)/Ab/CRP multilayers. Adapted from [<a href="#B88-sensors-19-03425" class="html-bibr">88</a>].</p> Full article ">Figure 7
<p>(<b>a</b>) Schematic structure of the floating-gate organic EDLT, in which the floating gate is modified with ssDNA probes and 6-mercaptohexanol (MCH), and then used to capture complementary DNA. (<b>b</b>) The sensitivity in terms of transfer curves shift (ΔV) as a function of the concentration of complementary DNA. The response is well fit by a Langmuir isotherm. The background level is defined as the sensor response to random DNA. (<b>c</b>) The sensor responses to complementary DNA, three kinds of mismatched DNA, and random DNA sequence, respectively. Adapted from [<a href="#B92-sensors-19-03425" class="html-bibr">92</a>].</p> Full article ">Figure 8
<p>(<b>a</b>) Sketch map of the chitosan-gated IZO-EDLT sensor. (<b>b</b>) Leakage curve of the chitosan electrolyte film. (<b>c</b>) The sensitivity of the chitosan-gate EDLT as a function of the gate sweep rate. The red dash line indicates the Nernst limit. (<b>d</b>) Stability test driven by square-wave gate voltages of −1.5 V to 1.5 V applied to the pH 7 solution at V<sub>DS</sub> = 0.1 V. Adapted from [<a href="#B109-sensors-19-03425" class="html-bibr">109</a>].</p> Full article ">Figure 9
<p>(<b>a</b>) The schematic diagram of the laterally coupled dual-gate IZO-based EDLT sensor. (<b>b</b>) The specific capacitance vs frequency curves of the SiO<sub>2</sub> electrolyte films for metal gate and solution gate (pH = 6.0), respectively. Inset: the lateral gate test structure. (<b>c</b>) pH-dependent transfer characteristics of the laterally coupled IZO-EDLT in single-gate sensing mode. (<b>d</b>) pH-dependent transfer characteristics of the laterally coupled IZO-EDLT in dual-gate sensing mode at different V<sub>REF</sub> of −0.6, 0, and 0.6 V, respectively. Adapted from [<a href="#B113-sensors-19-03425" class="html-bibr">113</a>].</p> Full article ">Figure 10
<p>(<b>a</b>) Schematic measurement setup of the solution-gated CNT transistor sensing platform. (<b>b</b>) Conductance of CNT vs V<sub>Pt</sub> at different stages. (<b>c</b>) Conductance recordings of DNA probe modified CNT transistor immersed in PBS without complementary DNA target. (<b>d</b>) Conductance-based histograms of time intervals extracted from the data in (<b>c</b>). (<b>e</b>) Conductance recordings of DNA probe modified CNT transistor immersed in PBS with 1μM complementary DNA target. (<b>f</b>) Conductance-based histograms of time intervals extracted from the data in (<b>d</b>). The two levels are fitted by Gaussian distributions. Adapted from [<a href="#B132-sensors-19-03425" class="html-bibr">132</a>].</p> Full article ">Figure 11
<p>(<b>a</b>) Schematic diagram of an electrolyte-gated graphene transistor operated in a frequency-doubling mode. In sensing process, the hybridization of negatively charged ssDNA with peptide nucleic acid (PNA) anchored on graphene surface occurs. (<b>b</b>) Transfer characteristics of electrolyte-gated graphene transistors before and after ssDNA adsorption in 1 mM PBS or in 100 mM KCl solutions, respectively. (<b>c</b>) Responses of A<sub>out</sub>-V<sub>ref</sub> characteristics for electrolyte-gated graphene transistors to the ssDNA adsorption in 1 mM PBS. Recordings of A<sub>out</sub> in real time upon the addition of (<b>d</b>) 1 nM and (<b>e</b>) 100 pM ssDNA into 1 mM PBS, respectively. Adapted from [<a href="#B143-sensors-19-03425" class="html-bibr">143</a>].</p> Full article ">Figure 12
<p>(<b>a</b>) Scheme map of spike sensing measurements on an oxide-based EDLT device. (<b>b</b>) The spike response to different solutions with pH value increasing from 4 to 10. (<b>c</b>) The logarithm of current peak as a function of solution pH value. (<b>d</b>) pH-dependent energy dissipation in single-spike sensing mode. (<b>e</b>) The variations in sensitivity and (<b>f</b>) energy dissipation (pH = 10) against different gate bias applied on pH solutions. Adapted from [<a href="#B152-sensors-19-03425" class="html-bibr">152</a>].</p> Full article ">Figure 13
<p>(<b>a</b>) Sketch map of IZO-based synaptic EDLTs. (<b>b</b>) Channel current responses to voltage spike (1 V, 10 ms) under different pH solutions. Inset: the enlarged view of current decay curve at pH = 4, which is well fitted with a stretched exponential function. (<b>c</b>) Response of EPSC to two successive presynaptic spikes (1 V, 10 ms) with a time interval of 20 ms at pH = 4. (<b>d</b>) PPF ratios as a function of time intervals under different pH solutions. (<b>e</b>) EPSCs amplitude gains (A<sub>n</sub>/A<sub>1</sub>) against pH values measured with different spike numbers. Inset: the response of EPSC to five successive V<sub>GS</sub> spikes (1 V, 10 ms) at pH = 4. (<b>f</b>) The relative amplitude gains (Gain<sub>pH4</sub>/Gain<sub>pH10</sub>) between pH = 4 and pH = 10 as a function of training times. Adapted from [<a href="#B153-sensors-19-03425" class="html-bibr">153</a>].</p> Full article ">
14 pages, 1333 KiB  
Article
Artificial Neural Networks for Forecasting Passenger Flows on Metro Lines
by Mariano Gallo, Giuseppina De Luca, Luca D’Acierno and Marilisa Botte
Sensors 2019, 19(15), 3424; https://doi.org/10.3390/s19153424 - 5 Aug 2019
Cited by 44 | Viewed by 5336
Abstract
Forecasting user flows on transportation networks is a fundamental task for Intelligent Transport Systems (ITSs). Indeed, most control and management strategies on transportation systems are based on the knowledge of user flows. For implementing ITS strategies, the forecast of user flows on some [...] Read more.
Forecasting user flows on transportation networks is a fundamental task for Intelligent Transport Systems (ITSs). Indeed, most control and management strategies on transportation systems are based on the knowledge of user flows. For implementing ITS strategies, the forecast of user flows on some network links obtained as a function of user flows on other links (for instance, where data are available in real time with sensors) may provide a significant contribution. In this paper, we propose the use of Artificial Neural Networks (ANNs) for forecasting metro onboard passenger flows as a function of passenger counts at station turnstiles. We assume that metro station turnstiles record the number of passengers entering by means of an automatic counting system and that these data are available every few minutes (temporal aggregation); the objective is to estimate onboard passengers on each track section of the line (i.e., between two successive stations) as a function of turnstile data collected in the previous periods. The choice of the period length may depend on service schedules. Artificial Neural Networks are trained by using simulation data obtained with a dynamic loading procedure of the rail line. The proposed approach is tested on a real-scale case: Line 1 of the Naples metro system (Italy). Numerical results show that the proposed approach is able to forecast the flows on metro sections with satisfactory precision. Full article
(This article belongs to the Special Issue Intelligent Transportation Related Complex Systems and Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>Types of metro stations: (<b>a</b>) turnstiles at the entrance; (<b>b</b>) turnstiles at platform accesses.</p> Full article ">Figure 2
<p>Line 1 route.</p> Full article ">Figure 3
<p>Dispersion diagrams ANN a_1_20 (best, upper diagram; worst, lower diagram).</p> Full article ">Figure 3 Cont.
<p>Dispersion diagrams ANN a_1_20 (best, upper diagram; worst, lower diagram).</p> Full article ">Figure 4
<p>Dispersion diagrams ANN b_2_10 (best, upper diagram; worst, lower diagram).</p> Full article ">Figure 4 Cont.
<p>Dispersion diagrams ANN b_2_10 (best, upper diagram; worst, lower diagram).</p> Full article ">
20 pages, 3936 KiB  
Article
Intelligent Rapid Adaptive Offloading Algorithm for Computational Services in Dynamic Internet of Things System
by Xuejing Li, Yajuan Qin, Huachun Zhou, Yongtao Cheng, Zhewei Zhang and Zhengyang Ai
Sensors 2019, 19(15), 3423; https://doi.org/10.3390/s19153423 - 4 Aug 2019
Cited by 8 | Viewed by 3708
Abstract
As restricted resources have seriously limited the computational performance of massive Internet of things (IoT) devices, better processing capability is urgently required. As an innovative technology, multi-access edge computing can provide cloudlet capabilities by offloading computation-intensive services from devices to a nearby edge [...] Read more.
As restricted resources have seriously limited the computational performance of massive Internet of things (IoT) devices, better processing capability is urgently required. As an innovative technology, multi-access edge computing can provide cloudlet capabilities by offloading computation-intensive services from devices to a nearby edge server. This paper proposes an intelligent rapid adaptive offloading (IRAO) algorithm for a dynamic IoT system to increase overall computational performance and simultaneously keep the fairness of multiple participants, which can achieve agile centralized control and solve the joint optimization problems related to offloading policy and resource allocation. For reducing algorithm execution time, we apply machine learning methods and construct an adaptive learning-based framework consisting of offloading decision-making, radio resource slicing and algorithm parameters updating. In particular, the offloading policy can be rapidly derived from an estimation algorithm based on a deep neural network, which uses an experience replay training method to improve model accuracy and adopts an asynchronous sampling trick to enhance training convergence performance. Extensive simulations with different parameters are conducted to maintain the trade-off between accuracy and efficiency of the IRAO algorithm. Compared with other candidates, the results illustrate that the IRAO algorithm can achieve superior performance in terms of scalability, effectiveness and efficiency. Full article
(This article belongs to the Special Issue Intelligent Systems in Sensor Networks and Internet of Things)
►▼ Show Figures

Figure 1

Figure 1
<p>The system model with multiple devices and dynamic channels.</p> Full article ">Figure 2
<p>Iterated learning-based framework of intelligent rapid adaptive offloading (IRAO).</p> Full article ">Figure 3
<p>Training process of asynchronously trained deep neural network (AsyDNN) model.</p> Full article ">Figure 4
<p>Frequency distribution histogram of different channel power gains.</p> Full article ">Figure 5
<p>Designed architecture of DNN.</p> Full article ">Figure 6
<p>The IRAO training processes of different learning rates.</p> Full article ">Figure 7
<p>Training processes and training losses of the IRAO algorithm with different DNN models: (<b>a</b>) the training processes under 10 users; (<b>b</b>) the training processes under 20 users; (<b>c</b>) the training losses under 10 users; (<b>d</b>) the training losses under 20 users.</p> Full article ">Figure 8
<p>Scalability analyses of training processes for the IRAO algorithm.</p> Full article ">Figure 9
<p>Algorithm effectiveness comparison of the IRAO and candidates.</p> Full article ">Figure 10
<p>Algorithm efficiency evaluation of IRAO and candidates.</p> Full article ">
16 pages, 2601 KiB  
Article
Machine Learning Classification for Assessing the Degree of Stenosis and Blood Flow Volume at Arteriovenous Fistulas of Hemodialysis Patients Using a New Photoplethysmography Sensor Device
by Pei-Yu Chiang, Paul C. -P. Chao, Tse-Yi Tu, Yung-Hua Kao, Chih-Yu Yang, Der-Cherng Tarng and Chin-Long Wey
Sensors 2019, 19(15), 3422; https://doi.org/10.3390/s19153422 - 4 Aug 2019
Cited by 11 | Viewed by 5880
Abstract
The classifier of support vector machine (SVM) learning for assessing the quality of arteriovenous fistulae (AVFs) in hemodialysis (HD) patients using a new photoplethysmography (PPG) sensor device is presented in this work. In clinical practice, there are two important indices for assessing the [...] Read more.
The classifier of support vector machine (SVM) learning for assessing the quality of arteriovenous fistulae (AVFs) in hemodialysis (HD) patients using a new photoplethysmography (PPG) sensor device is presented in this work. In clinical practice, there are two important indices for assessing the quality of AVF: the blood flow volume (BFV) and the degree of stenosis (DOS). In hospitals, the BFV and DOS of AVFs are nowadays assessed using an ultrasound Doppler machine, which is bulky, expensive, hard to use, and time consuming. In this study, a newly-developed PPG sensor device was utilized to provide patients and doctors with an inexpensive and small-sized solution for ubiquitous AVF assessment. The readout in this sensor was custom-designed to increase the signal-to-noise ratio (SNR) and reduce the environment interference via maximizing successfully the full dynamic range of measured PPG entering an analog–digital converter (ADC) and effective filtering techniques. With quality PPG measurements obtained, machine learning classifiers including SVM were adopted to assess AVF quality, where the input features are determined based on optical Beer–Lambert’s law and hemodynamic model, to ensure all the necessary features are considered. Finally, the clinical experiment results showed that the proposed PPG sensor device successfully achieved an accuracy of 87.84% based on SVM analysis in assessing DOS at AVF, while an accuracy of 88.61% was achieved for assessing BFV at AVF. Full article
(This article belongs to the Special Issue Non-Invasive Biomedical Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>A typical photoplethysmography (PPG) signal, which consists of a very large DC component (&gt;90%) and a very small AC component (&lt;10%) with pulse frequency the same as heart rate (about 50–110 bpm).</p> Full article ">Figure 2
<p>The coordinate axis used in this work, where the <span class="html-italic">z</span>-axis lies along the blood vessel with the measuring spot assumed to be at <span class="html-italic">z</span> = 0.</p> Full article ">Figure 3
<p>The definition of degree of stenosis (DOS), which is defined as the ratio of the stenosis to non-stenosis areas, where <span class="html-italic">h</span><sub>1</sub> and <span class="html-italic">h</span><sub>2</sub> denote the wall thicknesses of normal vessel and the vessel with endothelia cell proliferation.</p> Full article ">Figure 4
<p>The system architecture of the proposed wireless PPG sensor system for accessing arteriovenous fistulas (AVFs), which is composed of a 904 nm wavelength light-emitting diode (LED), a photodiode (PD), a transimpedance amplifier (TIA), a band-pass filter, an analog–digital converter (ADC), a microcontroller unit (MCU), a wireless communication interface, and the proposed classifiers for accessing DOS and blood flow volume (BFV) of AVFs.</p> Full article ">Figure 5
<p>A photo of the proposed portable, wireless, small-sized PPG sensor device for assessing AVF quality.</p> Full article ">Figure 6
<p>Measurement by the PPG sensor.</p> Full article ">Figure 7
<p>A typical experimental PPG waveform measured by the proposed PPG sensor.</p> Full article ">
9 pages, 2985 KiB  
Article
An Event Recognition Method for Φ-OTDR Sensing System Based on Deep Learning
by Yi Shi, Yuanye Wang, Lei Zhao and Zhun Fan
Sensors 2019, 19(15), 3421; https://doi.org/10.3390/s19153421 - 4 Aug 2019
Cited by 105 | Viewed by 6924
Abstract
Phase-sensitive optical time domain reflectometer (Φ-OTDR) based distributed optical fiber sensing system has been widely used in many fields such as long range pipeline pre-warning, perimeter security and structure health monitoring. However, the lack of event recognition ability is always being the bottleneck [...] Read more.
Phase-sensitive optical time domain reflectometer (Φ-OTDR) based distributed optical fiber sensing system has been widely used in many fields such as long range pipeline pre-warning, perimeter security and structure health monitoring. However, the lack of event recognition ability is always being the bottleneck of Φ-OTDR in field application. An event recognition method based on deep learning is proposed in this paper. This method directly uses the temporal-spatial data matrix from Φ-OTDR as the input of a convolutional neural network (CNN). Only a simple bandpass filtering and a gray scale transformation are needed as the pre-processing, which achieves real-time. Besides, an optimized network structure with small size, high training speed and high classification accuracy is built. Experiment results based on 5644 events samples show that this network can achieve 96.67% classification accuracy in recognition of 5 kinds of events and the retraining time is only 7 min for a new sensing setup. Full article
(This article belongs to the Special Issue Fiber-Based Sensing Technology: Recent Progresses and New Challenges)
►▼ Show Figures

Figure 1

Figure 1
<p>The distributed optical fiber sensing system.</p> Full article ">Figure 2
<p>The data matrix after bandpass filter.</p> Full article ">Figure 3
<p>The typical gray image of each event type.</p> Full article ">Figure 4
<p>The relationship between model size and classification accuracy.</p> Full article ">Figure 5
<p>The optimized network structure (the red cube denotes convolution operation and the blue cube denotes pooling operation).</p> Full article ">Figure 6
<p>The learning curve of training.</p> Full article ">Figure 7
<p>Loss curve (<b>a</b>) and classification accuracy curve (<b>b</b>) of training.</p> Full article ">Figure 8
<p>Confusion matrix of five events’ classification.</p> Full article ">Figure 9
<p>Accuracy curve of optimized network (green) and Inception-v3 (red).</p> Full article ">
17 pages, 397 KiB  
Article
Blood Pressure Estimation from Photoplethysmogram Using a Spectro-Temporal Deep Neural Network
by Gašper Slapničar, Nejc Mlakar and Mitja Luštrek
Sensors 2019, 19(15), 3420; https://doi.org/10.3390/s19153420 - 4 Aug 2019
Cited by 229 | Viewed by 16263
Abstract
Blood pressure (BP) is a direct indicator of hypertension, a dangerous and potentially deadly condition. Regular monitoring of BP is thus important, but many people have aversion towards cuff-based devices, and their limitation is that they can only be used at rest. Using [...] Read more.
Blood pressure (BP) is a direct indicator of hypertension, a dangerous and potentially deadly condition. Regular monitoring of BP is thus important, but many people have aversion towards cuff-based devices, and their limitation is that they can only be used at rest. Using just a photoplethysmogram (PPG) to estimate BP is a potential solution investigated in our study. We analyzed the MIMIC III database for high-quality PPG and arterial BP waveforms, resulting in over 700 h of signals after preprocessing, belonging to 510 subjects. We then used the PPG alongside its first and second derivative as inputs into a novel spectro-temporal deep neural network with residual connections. We have shown in a leave-one-subject-out experiment that the network is able to model the dependency between PPG and BP, achieving mean absolute errors of 9.43 for systolic and 6.88 for diastolic BP. Additionally we have shown that personalization of models is important and substantially improves the results, while deriving a good general predictive model is difficult. We have made crucial parts of our study, especially the list of used subjects and our neural network code, publicly available, in an effort to provide a solid baseline and simplify potential comparison between future studies on an explicit MIMIC III subset. Full article
(This article belongs to the Section Biosensors)
►▼ Show Figures

Figure 1

Figure 1
<p>Ideal photoplethysmogram (PPG) cycle waveform and its first and second derivatives next to distorted waveforms. The ideal example has a single large systolic peak and a single lower diastolic peak afterwards, while the anomalies have too many or too few peaks. All data is taken from the MIMIC III database [<a href="#B11-sensors-19-03420" class="html-bibr">11</a>].</p> Full article ">Figure 2
<p>Flat lines anomaly can be observed in the arterial blood pressure (ABP) signal.</p> Full article ">Figure 3
<p>Flat peaks anomaly can be observed in the ABP signal.</p> Full article ">Figure 4
<p>Schematic pipeline of our system.</p> Full article ">Figure 5
<p>Distributions of systolic blood pressure (SBP) and distolic blood pressure (DBP) in our final data.</p> Full article ">Figure 6
<p>Schematic showing of our neural network architecture.</p> Full article ">
19 pages, 7782 KiB  
Article
Investigation on Vortex-Induced Vibration Experiment of a Standing Variable-Tension Deepsea Riser Based on BFBG Sensor Technology
by Peng Li, Aijun Cong, Zhengkai Dong, Yu Wang, Yu Liu, Haiyan Guo, Xiaomin Li and Qiang Fu
Sensors 2019, 19(15), 3419; https://doi.org/10.3390/s19153419 - 4 Aug 2019
Cited by 6 | Viewed by 3587
Abstract
A vortex-induced vibration (VIV) experiment on a standing variable-tension deepsea riser was conducted to investigate the applicability and sensitivity of Bare Fiber Bragg Grating (BFBG) sensor technology for testing deepsea riser vibrations. The dominant frequencies, dimensionless displacements, in-line and cross-flow couplings of the [...] Read more.
A vortex-induced vibration (VIV) experiment on a standing variable-tension deepsea riser was conducted to investigate the applicability and sensitivity of Bare Fiber Bragg Grating (BFBG) sensor technology for testing deepsea riser vibrations. The dominant frequencies, dimensionless displacements, in-line and cross-flow couplings of the riser VIV under different top tensions were observed through wavelet transform and modal decomposition. The result indicated that, excited by the same external flow velocities, the cross-flow and in-line dominant frequencies of the riser both decreased with increasing top tension. In terms of displacement responses, increasing top tension caused the root mean square (RMS) displacement to decrease and the vibration amplitude to reduce. In terms of cross-flow and in-line coupling, the closer a location is to the ends of the riser, the smaller the trajectory is and the more standard the “8” becomes. During top tension increases, there exists a “lag” in the time when the riser’s vibration trajectory becomes an “8”. The Slalom Surround Installation approach can effectively prevent the local breakage of the optical fiber string. BFBG sensor technology can give an accurate presentation of the displacement time history, vibration amplitude and frequency of the riser, provides a clear picture of how the riser’s mode and VIV evolve as a function of flow velocity. Full article
(This article belongs to the Section Physical Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>Schematic diagram of the operation and Select configuration diagram of the FBG sensor. (<b>a</b>) Working principle of Fiber Bragg grating (FBG); (<b>b</b>) wavelength division multiplexing (WDM) technology for FBG sensors; (<b>c</b>) Selection and configuration of FBG sensors.</p> Full article ">Figure 2
<p>General setup of the experiment (<b>a</b>) Top view; (<b>b</b>) Elevation view; (<b>c</b>) Side view.</p> Full article ">Figure 3
<p>Actual working conditions of installation and launching of riser and supporting structure.</p> Full article ">Figure 4
<p>Experimental model and mechanical properties test.</p> Full article ">Figure 5
<p>Layout of the top tension system and FBG sensors. (<b>a</b>) Riser model; (<b>b</b>) Control diagram of the top tension system; (<b>c</b>) FBG demodulator; (<b>d</b>) Bare fiber string and armored optical fiber; (<b>e</b>) Layout of FBG measuring points; (<b>f</b>) Installation of bare fibers.</p> Full article ">Figure 6
<p>The dimensionless dominant frequency of VIV in the direction of cross-flow (<b>a</b>) and in-line (<b>b</b>) varies with reduced velocity and Strouhal number fitting under different top tension.</p> Full article ">Figure 7
<p>The dimensionless dominant frequency of VIV in the direction of cross-flow varies with the outflow velocity under different top tension.</p> Full article ">Figure 8
<p>Strain time history (top left), enlargements of strain time history (top right), wavelet time–frequency scale diagrams (bottom left) and power spectral density diagrams (bottom right) of cross-flow riser VIV versus flow velocity under different top tensions (for each level of top tension, (<b>a</b>)–(<b>d</b>), (<b>e</b>)–(<b>h</b>), (<b>i</b>)–(<b>l</b>), (<b>m</b>)–(<b>p</b>), (<b>q</b>)–(<b>t</b>), correspond to four flow velocities: 0.35, 0.45, 0.55, and 0.6 m/s).</p> Full article ">Figure 8 Cont.
<p>Strain time history (top left), enlargements of strain time history (top right), wavelet time–frequency scale diagrams (bottom left) and power spectral density diagrams (bottom right) of cross-flow riser VIV versus flow velocity under different top tensions (for each level of top tension, (<b>a</b>)–(<b>d</b>), (<b>e</b>)–(<b>h</b>), (<b>i</b>)–(<b>l</b>), (<b>m</b>)–(<b>p</b>), (<b>q</b>)–(<b>t</b>), correspond to four flow velocities: 0.35, 0.45, 0.55, and 0.6 m/s).</p> Full article ">Figure 9
<p>Cross-flow and in-line riser RMS dimensionless displacements versus flow velocity under different top tensions.</p> Full article ">Figure 10
<p>Each (1/20) s of the riser forms a cross-flow deflection curve in motion, <span class="html-italic">T</span> = 19.8 N, <span class="html-italic">U</span> = 0.3 m/s, <span class="html-italic">U</span> = 0.45 m/s, <span class="html-italic">U</span> = 0.6 m/s.</p> Full article ">Figure 11
<p>Cross-flow and in-line riser RMS displacements curves along the length of the riser at U = 0.3 m/s (<b>a</b>), U = 0.45 m/s (<b>b</b>), U = 0.6 m/s (<b>c</b>) under different top tensions.</p> Full article ">Figure 12
<p>Time-varying graphs of riser dimensionless amplitudes at U = 0.3 m/s, U = 0.45 m/s, U = 0.6 m/s under different top tensions. (<b>a1</b>–<b>a5</b>) <span class="html-italic">T</span> = 19.8–98.0 N, <span class="html-italic">U</span> = 0.3m/s, dimensionless amplitude time-varying diagram of riser; (<b>b1</b>–<b>b5</b>) <span class="html-italic">T</span> = 19.8–98.0 N, <span class="html-italic">U</span> = 0.45 m/s, dimensionless amplitude time-varying diagram of riser; (<b>c1</b>–<b>c5</b>) <span class="html-italic">T</span> = 19.8–98.0 N, <span class="html-italic">U</span> = 0.6 m/s, dimensionless amplitude time-varying diagram of riser.</p> Full article ">Figure 13
<p>Riser vibration trajectories versus flow velocity under different top tensions.</p> Full article ">Figure 14
<p>Three-dimensional riser axial motion trajectories at <span class="html-italic">U</span> = 0.45 m/s under different top tensions.</p> Full article ">
21 pages, 2398 KiB  
Article
A Hybrid Sliding Window Optimizer for Tightly-Coupled Vision-Aided Inertial Navigation System
by Junxiang Jiang, Xiaoji Niu, Ruonan Guo and Jingnan Liu
Sensors 2019, 19(15), 3418; https://doi.org/10.3390/s19153418 - 4 Aug 2019
Cited by 8 | Viewed by 3904
Abstract
The fusion of visual and inertial measurements for motion tracking has become prevalent in the robotic community, due to its complementary sensing characteristics, low cost, and small space requirements. This fusion task is known as the vision-aided inertial navigation system problem. We present [...] Read more.
The fusion of visual and inertial measurements for motion tracking has become prevalent in the robotic community, due to its complementary sensing characteristics, low cost, and small space requirements. This fusion task is known as the vision-aided inertial navigation system problem. We present a novel hybrid sliding window optimizer to achieve information fusion for a tightly-coupled vision-aided inertial navigation system. It possesses the advantages of both the conditioning-based method and the prior-based method. A novel distributed marginalization method was also designed based on the multi-state constraints method with significant efficiency improvement over the traditional method. The performance of the proposed algorithm was evaluated with the publicly available EuRoC datasets and showed competitive results compared with existing algorithms. Full article
(This article belongs to the Special Issue Multi-Sensor Systems for Positioning and Navigation)
►▼ Show Figures

Figure 1

Figure 1
<p>The structure of the hybrid sliding window optimizer.</p> Full article ">Figure 2
<p>Results for the <span class="html-italic">MH_03_medium</span> image sequence after visual-inertial initialization: (<b>a</b>) Full trajectory after visual-inertial initialization; (<b>b</b>) Translation deviation with respect to the ground truth; (<b>c</b>) Rotation deviation of camera-to-IMU transformation; (<b>d</b>) Translation deviation of camera-to-IMU transformation.</p> Full article ">Figure 3
<p>Results for the <span class="html-italic">V1_03_difficult</span> image sequence after visual-inertial initialization: (<b>a</b>) Full trajectory after visual-inertial initialization; (<b>b</b>) Translation deviation with respect to the ground truth; (<b>c</b>) Rotation deviation of camera-to-IMU transformation; (<b>d</b>) Translation deviation of camera-to-IMU transformation.</p> Full article ">Figure 4
<p>The yaw error of our implementation on Machine Hall datasets.</p> Full article ">Figure 5
<p>The yaw error of our implementation on Vicon room datasets.</p> Full article ">Figure 6
<p>The comparison of the time consumption of our method and the traditional method.</p> Full article ">
18 pages, 2606 KiB  
Article
A Novel Method for Soil Organic Matter Determination by Using an Artificial Olfactory System
by Longtu Zhu, Honglei Jia, Yibing Chen, Qi Wang, Mingwei Li, Dongyan Huang and Yunlong Bai
Sensors 2019, 19(15), 3417; https://doi.org/10.3390/s19153417 - 4 Aug 2019
Cited by 25 | Viewed by 5541
Abstract
Soil organic matter (SOM) is a major indicator of soil fertility and nutrients. In this study, a soil organic matter measuring method based on an artificial olfactory system (AOS) was designed. An array composed of 10 identical gas sensors controlled at different temperatures [...] Read more.
Soil organic matter (SOM) is a major indicator of soil fertility and nutrients. In this study, a soil organic matter measuring method based on an artificial olfactory system (AOS) was designed. An array composed of 10 identical gas sensors controlled at different temperatures was used to collect soil gases. From the response curve of each sensor, four features were extracted (maximum value, mean differential coefficient value, response area value, and the transient value at the 20th second). Then, soil organic matter regression prediction models were built based on back-propagation neural network (BPNN), support vector regression (SVR), and partial least squares regression (PLSR). The prediction performance of each model was evaluated using the coefficient of determination (R2), root-mean-square error (RMSE), and the ratio of performance to deviation (RPD). It was found that the R2 values between prediction (from BPNN, SVR, and PLSR) and observation were 0.880, 0.895, and 0.808. RMSEs were 14.916, 14.094, and 18.890, and RPDs were 2.837, 3.003, and 2.240, respectively. SVR had higher prediction ability than BPNN and PLSR and can be used to accurately predict organic matter contents. Thus, our findings offer brand new methods for predicting SOM. Full article
(This article belongs to the Section Chemical Sensors)
►▼ Show Figures

Figure 1

Figure 1
<p>The study area and sampling sites.</p> Full article ">Figure 2
<p>Artificial olfactory measurement setup.</p> Full article ">Figure 3
<p>Sensor circuit: (<b>a</b>) The basic measuring circuit of sensors; (<b>b</b>) temperature modulation circuit.</p> Full article ">Figure 4
<p>Response curves of the sensors: (<b>a</b>) Helium; (<b>b</b>) air; (<b>c</b>) soil gas.</p> Full article ">Figure 5
<p>Sensor array signals of soil samples: (<b>a</b>) Soil organic matter (SOM) content 12.19 mg/kg; (<b>b</b>) SOM content 23.11 mg/kg; (<b>c</b>) SOM content 48.79 mg/kg.</p> Full article ">Figure 6
<p>Back-propagation neural network <b>(</b>BPNN) predicted values and observed values of SOM: (<b>a</b>) Training set; (<b>b</b>) validation set.</p> Full article ">Figure 7
<p>Support vector regression (SVR) parameters selection: (<b>a</b>) Contour of rough selection; (<b>b</b>) contour of precise selection. log<sub>2</sub>C: Logarithm of C with the bottom number 2; log<sub>2</sub>σ<sup>2</sup>: Logarithm of σ<sup>2</sup> with the bottom number 2.</p> Full article ">Figure 8
<p>Calibration results and prediction results of SVR model: (<b>a</b>) Calibration; (<b>b</b>) prediction.</p> Full article ">Figure 9
<p>Number of principal component factors (PCFs) in partial least squares regression (PLSR): (<b>a</b>) Root mean square error of cross-validation (RMSECV); (<b>b</b>) Akaike information criterion (AIC).</p> Full article ">Figure 10
<p>Calibration and prediction results with the PLSR model: (<b>a</b>) Calibration; (<b>b</b>) prediction.</p> Full article ">Figure 11
<p>Comparison of prediction results from different models.</p> Full article ">
18 pages, 9193 KiB  
Article
3D-Printed Multilayer Sensor Structure for Electrical Capacitance Tomography
by Aleksandra Kowalska, Robert Banasiak, Andrzej Romanowski and Dominik Sankowski
Sensors 2019, 19(15), 3416; https://doi.org/10.3390/s19153416 - 4 Aug 2019
Cited by 25 | Viewed by 4714
Abstract
Presently, Electrical Capacitance Tomography (ECT) is positioned as a relatively mature and inexpensive tool for the diagnosis of non-conductive industrial processes. For most industrial applications, a hand-made approach for an ECT sensor and its 3D extended structure fabrication is used. Moreover, a hand-made [...] Read more.
Presently, Electrical Capacitance Tomography (ECT) is positioned as a relatively mature and inexpensive tool for the diagnosis of non-conductive industrial processes. For most industrial applications, a hand-made approach for an ECT sensor and its 3D extended structure fabrication is used. Moreover, a hand-made procedure is often inaccurate, complicated, and time-consuming. Another drawback is that a hand-made ECT sensor’s geometrical parameters, mounting base profile thickness, and electrode array shape usually depends on the structure of industrial test objects, tanks, and containers available on the market. Most of the traditionally fabricated capacitance tomography sensors offer external measurements only with electrodes localized outside of the test object. Although internal measurement is possible, it is often difficult to implement. This leads to limited in-depth scanning abilities and poor sensitivity distribution of traditionally fabricated ECT sensors. In this work we propose, demonstrate, and validate experimentally a new 3D ECT sensor fabrication process. The proposed solution uses a computational workflow that incorporates both 3D computer modeling and 3D-printing techniques. Such a 3D-printed structure can be of any shape, and the electrode layout can be easily fitted to a broad range of industrial applications. A developed solution offers an internal measurement due to negligible thickness of sensor mount base profile. This paper analyses and compares measurement capabilities of a traditionally fabricated 3D ECT sensor with novel 3D-printed design. The authors compared two types of the 3D ECT sensors using experimental capacitance measurements for a set of low-contrast and high-contrast permittivity distribution phantoms. The comparison demonstrates advantages and benefits of using the new 3D-printed spatial capacitance sensor regarding the significant fabrication time reduction as well as the improvement of overall measurement accuracy and stability. Full article
(This article belongs to the Special Issue Selected Papers from the 9th World Congress on Industrial Process Tomography)
►▼ Show Figures

Figure 1

Figure 1
<p>3D ECT system and its components.</p> Full article ">Figure 2
<p>3D printed ECT sensor building workflow.</p> Full article ">Figure 3
<p>Experimental setup hardware: left—32-channel ET3 measurement hardware, right—Agilent E4980A with 64-channel multiplexer device.</p> Full article ">Figure 4
<p>Pictures of two experimental constructions of 3D ECT sensors under study.</p> Full article ">Figure 5
<p>A concept of horizontal and vertical internal screening for 3D ECT sensor structure.</p> Full article ">Figure 6
<p>Arrangement of tested low-contrast objects according to <a href="#sensors-19-03416-t001" class="html-table">Table 1</a>—from the leftmost Test<span class="html-italic">A</span>, Test<span class="html-italic">B</span>, Test<span class="html-italic">C</span>.</p> Full article ">Figure 7
<p>A set of ten phantoms used during high-contrast media measurements tests. The mounting stand had 5 holes. Three 10 mm of diameter holes were positioned along the sensor profile diameter at given positions: “P1” at x = 70 mm, y = 70 mm; “P2” at x = 70 mm, y = 40 mm; “P3” at x = 70 mm, y = 10 mm. Two additional 40 mm of diameter holes were positioned symmetrically at x1 = 110 mm and x2 = 40 mm for y = 70 mm. All the rods were parallel to sensor walls.</p> Full article ">Figure 8
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32)for Test<math display="inline"><semantics> <msub> <mi>A</mi> <mrow> <mi>i</mi> <mi>n</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 9
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mi>A</mi> <mrow> <mi>o</mi> <mi>u</mi> <mi>t</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 10
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mi>B</mi> <mrow> <mi>i</mi> <mi>n</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 11
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mi>B</mi> <mrow> <mi>o</mi> <mi>u</mi> <mi>t</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 12
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mi>C</mi> <mrow> <mi>i</mi> <mi>n</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 13
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mi>C</mi> <mrow> <mi>o</mi> <mi>u</mi> <mi>t</mi> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan and magenta lines indicate calibration limits (0;1).</p> Full article ">Figure 14
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <mrow> <mn>2</mn> <mi>x</mi> <mn>40</mn> </mrow> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 15
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>20</mn> <mrow> <mi>P</mi> <mn>1</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 16
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>20</mn> <mrow> <mi>P</mi> <mn>2</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 17
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>20</mn> <mrow> <mi>P</mi> <mn>3</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 18
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>15</mn> <mrow> <mi>P</mi> <mn>1</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 19
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>15</mn> <mrow> <mi>P</mi> <mn>2</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 20
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>15</mn> <mrow> <mi>P</mi> <mn>3</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 21
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>10</mn> <mrow> <mi>P</mi> <mn>1</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 22
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>10</mn> <mrow> <mi>P</mi> <mn>2</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">Figure 23
<p>The 1st electrode measurement cycle (S1: 1-&gt;32 and S2: 1-&gt;32) for Test<math display="inline"><semantics> <msub> <mn>10</mn> <mrow> <mi>P</mi> <mn>3</mn> </mrow> </msub> </semantics></math> and S1—blue line, S2—red line. Cyan line indicates lower calibration limit.</p> Full article ">
14 pages, 388 KiB  
Article
Magnetic Communication Using High-Sensitivity Magnetic Field Detectors
by Maurice Hott, Peter A. Hoeher and Sebastian F. Reinecke
Sensors 2019, 19(15), 3415; https://doi.org/10.3390/s19153415 - 4 Aug 2019
Cited by 25 | Viewed by 6050
Abstract
In this article, an innovative approach for magnetic data communication is presented. For this purpose, the receiver coil of a conventional magneto-inductive communication system is replaced by a high-sensitivity wideband magnetic field sensor. The results show decisive advantages offered by sensitive magnetic field [...] Read more.
In this article, an innovative approach for magnetic data communication is presented. For this purpose, the receiver coil of a conventional magneto-inductive communication system is replaced by a high-sensitivity wideband magnetic field sensor. The results show decisive advantages offered by sensitive magnetic field sensors, including a higher communication range for small receiver units. This approach supports numerous mobile applications where receiver size is limited, possibly in conjunction with multiple detectors. Numerical results are supported by a prototype implementation employing an anisotropic magneto-resistive sensor. Full article
(This article belongs to the Special Issue Magnetic Sensing Technology, Materials and Applications)
►▼ Show Figures

Figure 1

Figure 1
<p>Conventional coil-to-coil topology (<b>a</b>) and coil-to-AMR topology under investigation (<b>b</b>).</p> Full article ">Figure 2
<p>AMR sensor schema referring to [<a href="#B14-sensors-19-03415" class="html-bibr">14</a>].</p> Full article ">Figure 3
<p>Measured voltage at the receiver output for different current through the transmitter coil (<b>a</b>) and the AMR sensor sensitivity curve (<b>b</b>).</p> Full article ">Figure 4
<p>Measured noise signal for different bandwidths.</p> Full article ">Figure 5
<p>Transmission range for coil-to-coil and coil-to-AMR communication depending on the transmitter coil radius (<b>a</b>) and receiver coil radius (<b>b</b>). Parameters are taken from <a href="#sensors-19-03415-t002" class="html-table">Table 2</a>.</p> Full article ">Figure 6
<p>Transmission range for coil-to-coil and coil-to-AMR communication depending on the permeability (<b>a</b>) and the receiver coil radius for different core materials (<b>b</b>). Parameters are taken from <a href="#sensors-19-03415-t002" class="html-table">Table 2</a> and <a href="#sensors-19-03415-t003" class="html-table">Table 3</a>. The dashed lines are for coil-to-AMR transmission.</p> Full article ">Figure 7
<p>Transmission range as a function of the magnetic field detection limit of the detector. Parameters are taken from <a href="#sensors-19-03415-t002" class="html-table">Table 2</a>. The maximum transmission distance based on the prototype implementation detection threshold is marked for three different bandwidths. The transmission range can be increased by using a more sensitive type of detector.</p> Full article ">
17 pages, 3239 KiB  
Article
Effects of AlN and BCN Thin Film Multilayer Design on the Reaction Time of Ni/Ni-20Cr Thin Film Thermocouples on Thermally Sprayed Al2O3
by Wolfgang Tillmann, David Kokalj, Dominic Stangier, Volker Schöppner and Hatice Malatyali
Sensors 2019, 19(15), 3414; https://doi.org/10.3390/s19153414 - 3 Aug 2019
Cited by 7 | Viewed by 4364
Abstract
Thin film thermocouples are widely used for local temperature determinations of surfaces. However, depending on the environment in which they are used, thin film thermocouples need to be covered by a wear or oxidation resistant top layer. With regard to the utilization in [...] Read more.
Thin film thermocouples are widely used for local temperature determinations of surfaces. However, depending on the environment in which they are used, thin film thermocouples need to be covered by a wear or oxidation resistant top layer. With regard to the utilization in wide-slit nozzles for plastic extrusion, Ni/Ni-20Cr thin film thermocouples were manufactured using direct-current (DC) magnetron sputtering combined with Aluminiumnitride (AlN) and Boron-Carbonitride (BCN) thin films. On the one hand, the deposition parameters of the nitride layers were varied to affect the chemical composition and morphology of the AlN and BCN thin films. On the other hand, the position of the nitride layers (below the thermocouple, above the thermocouple, around the thermocouple) was changed. Both factors were investigated concerning the influence on the Seebeck coefficient and the reaction behaviour of the thermocouples. Therefore, the impact of the nitride thin films on the morphology, physical structure, crystallite size, electrical resistance and hardness of the Ni and Ni-20Cr thin films is analysed. The investigations reveal that the Seebeck coefficient is not affected by the different architectures of the thermocouples. Nevertheless, the reaction time of the thermocouples can be significantly improved by adding a thermal conductive top coat over the thin films, whereas the top coat should have a coarse structure and low nitrogen content. Full article
(This article belongs to the Section Sensor Materials)
►▼ Show Figures

Figure 1

Figure 1
<p>Multilayer design of Ni/NiCr thin film thermocouples with and without nitride top and bottom layers.</p> Full article ">Figure 2
<p>Drawing of the design of the thin film thermocouples with corresponding sizes.</p> Full article ">Figure 3
<p>Scanning electron microscope (SEM) images of the topography of the polished Al<sub>2</sub>O<sub>3</sub> substrate as well as the AlN and BCN thin films.</p> Full article ">Figure 4
<p>SEM images of the topography (<b>top</b>) and cross-section (<b>bottom</b>) of Ni-20Cr thin films, deposited on a polished Al<sub>2</sub>O<sub>3</sub> substrate as well as on AlN and BCN thin films.</p> Full article ">Figure 5
<p>SEM images of the topography (top) of the Ni thin films deposited on a polished Al<sub>2</sub>O<sub>3</sub> substrate as well as on AlN and BCN thin films.</p> Full article ">Figure 6
<p>X-ray diffraction (XRD) patterns of the (<b>a</b>) Ni and (<b>b</b>) Ni-20Cr thin films, deposited on a polished Al<sub>2</sub>O<sub>3</sub> substrate as well as on AlN and BCN thin films.</p> Full article ">Figure 7
<p>Influence of the substrate layer on the crystallite size and electrical resistance of Ni and Ni-20Cr thin films.</p> Full article ">Figure 8
<p>Influence of the substrate layer on the (<b>a</b>) hardness and (<b>b</b>) Young’s modulus of Ni and Ni-20Cr thin films.</p> Full article ">Figure 9
<p>Calibration curve for the Ni/Ni-20Cr thermocouple combined with AlN-1 layers.</p> Full article ">Figure 10
<p>Long-time experiment showing the stability of the reference thin film thermocouple.</p> Full article ">Figure 11
<p>Step function response of the Ni/Ni-20Cr thin film coated with different AlN-1 layer options and step function response of the reference thermocouple.</p> Full article ">Figure 12
<p>Reaction time of the coated thin film thermocouples and the industrial type K reference.</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-198
Previous Issue
Next Issue
Back to TopTop