Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956629
Fei Chang, Yunqiang Duan, Min Liu, Mingyu Dong
High-quality image acquisition under uncertain imaging conditions (such as uneven and varied illuminations, various viewpoints and different object distances, etc.) is a very challenging task. However, the imaging quality of industrial vision inspection system is vital to subsequent image processing, especially for those challenging detection tasks, such as tiny defect inspection of paint car-body surfaces. In order to overcome the challenge of image acquisition due to uncertain imaging conditions, a two-stage adaptive illumination adjustment method is proposed to handle the uncertainty caused by diversities of lighting, viewpoint and object distance. Our algorithm framework has been implemented and applied to the mobile inspection system deployed in a car painting factory for tiny defect detection of paint car-body surfaces. The efficiency and effectiveness of our method has been validated by the actual industrial application. As a result, the proposed coarse-to-fine framework can be viewed as an adaptive hard-adjustment solution for industrial vision inspection system under uncertain imaging conditions.
{"title":"Coarse-to-Fine Adaptive Illumination Hard-Adjustment for Vision Inspection System Under Uncertain Imaging Conditions","authors":"Fei Chang, Yunqiang Duan, Min Liu, Mingyu Dong","doi":"10.1109/SENSORS43011.2019.8956629","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956629","url":null,"abstract":"High-quality image acquisition under uncertain imaging conditions (such as uneven and varied illuminations, various viewpoints and different object distances, etc.) is a very challenging task. However, the imaging quality of industrial vision inspection system is vital to subsequent image processing, especially for those challenging detection tasks, such as tiny defect inspection of paint car-body surfaces. In order to overcome the challenge of image acquisition due to uncertain imaging conditions, a two-stage adaptive illumination adjustment method is proposed to handle the uncertainty caused by diversities of lighting, viewpoint and object distance. Our algorithm framework has been implemented and applied to the mobile inspection system deployed in a car painting factory for tiny defect detection of paint car-body surfaces. The efficiency and effectiveness of our method has been validated by the actual industrial application. As a result, the proposed coarse-to-fine framework can be viewed as an adaptive hard-adjustment solution for industrial vision inspection system under uncertain imaging conditions.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"44 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87333583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956801
Ozberk Ozturk, M. Yapici
In this study, we report, for the first time, wearable graphene textile electrodes for monitoring of muscular activity and surface electromyography (sEMG) applications. The feasibility of graphene textiles in wearable muscular monitoring was successfully demonstrated by the acquisition of sEMG signals with wearable graphene textiles, and their performance was benchmarked against commercial, wet Ag/AgCl electrodes. Comparisons were performed in terms of signal-to-noise ratio (SNR), cross correlation and sensitivity to power-line interference. Despite their larger susceptibility to power line interference, graphene textile electrodes displayed excellent similarity with Ag/AgCl electrodes in terms of signal-to-noise ratio (SNR) and signal morphology; with correlation values reaching up to 97 % for sEMG signals acquired from the biceps brachii muscle.
{"title":"Muscular Activity Monitoring and Surface Electromyography (sEMG) with Graphene Textiles","authors":"Ozberk Ozturk, M. Yapici","doi":"10.1109/SENSORS43011.2019.8956801","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956801","url":null,"abstract":"In this study, we report, for the first time, wearable graphene textile electrodes for monitoring of muscular activity and surface electromyography (sEMG) applications. The feasibility of graphene textiles in wearable muscular monitoring was successfully demonstrated by the acquisition of sEMG signals with wearable graphene textiles, and their performance was benchmarked against commercial, wet Ag/AgCl electrodes. Comparisons were performed in terms of signal-to-noise ratio (SNR), cross correlation and sensitivity to power-line interference. Despite their larger susceptibility to power line interference, graphene textile electrodes displayed excellent similarity with Ag/AgCl electrodes in terms of signal-to-noise ratio (SNR) and signal morphology; with correlation values reaching up to 97 % for sEMG signals acquired from the biceps brachii muscle.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"24 11 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91052665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956734
Reza Nouri, Zifan Tang, W. Guan
Nanopore sensor conceptually represents an ideal single molecule counting device due to its unique partitioning-free, label-free electronic sensing. Existing theories and experiments have shown that sample concentration is proportional to the molecule translocation rate. However, a detailed nanopore geometry and size characterization or a calibration curve of concentration standards are often required for quantifying the unknown sample. In this work, we proposed and validated a calibration-free nanopore single molecule digital counting method for isolated molecule quantification. With the background ions as the in-situ references, the molecule translocation rates can be normalized to the ion translocation rates (baseline current). This in-situ reference alleviates the requirement for knowing the nanopore geometry and size or generating a calibration curve. In recognition of this effect, we developed a quantitative model for molecule quantification without the need for prior knowledge of experimental conditions such as nanopore geometry, size, and applied voltage. This model was experimentally validated for different nanopores and DNA molecules with different sizes. We anticipate this calibration-free digital counting approach would provide a new avenue for nanopore-based molecule sensing.
{"title":"Calibration-Free Electrical Quantification of Single Molecules Using Nanopore Digital Counting","authors":"Reza Nouri, Zifan Tang, W. Guan","doi":"10.1109/SENSORS43011.2019.8956734","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956734","url":null,"abstract":"Nanopore sensor conceptually represents an ideal single molecule counting device due to its unique partitioning-free, label-free electronic sensing. Existing theories and experiments have shown that sample concentration is proportional to the molecule translocation rate. However, a detailed nanopore geometry and size characterization or a calibration curve of concentration standards are often required for quantifying the unknown sample. In this work, we proposed and validated a calibration-free nanopore single molecule digital counting method for isolated molecule quantification. With the background ions as the in-situ references, the molecule translocation rates can be normalized to the ion translocation rates (baseline current). This in-situ reference alleviates the requirement for knowing the nanopore geometry and size or generating a calibration curve. In recognition of this effect, we developed a quantitative model for molecule quantification without the need for prior knowledge of experimental conditions such as nanopore geometry, size, and applied voltage. This model was experimentally validated for different nanopores and DNA molecules with different sizes. We anticipate this calibration-free digital counting approach would provide a new avenue for nanopore-based molecule sensing.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"77 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91096656","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956491
Debnath Maji, S. Pourang, U. Sekhon, A. Gupta, M. Suster, P. Mohseni
This paper reports on capturing the morphological changes in red blood cells (RBCs) during platelet-driven blood clot retraction using a microfluidic dielectric sensor, termed ClotChip. The sensor is based on the technique of dielectric spectroscopy and incorporates screen-printed, gold, sensing and floating electrodes embedded into a microfluidic channel with a total sample volume of < 10μL to form a 3D, parallel-plate, capacitive sensing area. The ClotChip readout parameter, Δεr,max, is shown to be sensitive to the degree of deformation in the shape of RBCs caused by either impairing the contractile forces acting on the RBCs via a depletion of activated platelets or altering their membrane rigidity. Furthermore, a baseline range for Δεr,max parameter is established using 95 healthy whole blood samples, and the ClotChip is shown to detect weak clot formation, when the platelet count drops below 15,000 per μL. This work highlights the potential of ClotChip as a platform technology for rapid assessment of platelet loss during traumatic hemorrhage at the point-of-injury.
{"title":"Toward Diagnosis of Platelet Loss in Trauma Injury Using a Microfluidic Dielectric Sensor","authors":"Debnath Maji, S. Pourang, U. Sekhon, A. Gupta, M. Suster, P. Mohseni","doi":"10.1109/SENSORS43011.2019.8956491","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956491","url":null,"abstract":"This paper reports on capturing the morphological changes in red blood cells (RBCs) during platelet-driven blood clot retraction using a microfluidic dielectric sensor, termed ClotChip. The sensor is based on the technique of dielectric spectroscopy and incorporates screen-printed, gold, sensing and floating electrodes embedded into a microfluidic channel with a total sample volume of < 10μL to form a 3D, parallel-plate, capacitive sensing area. The ClotChip readout parameter, Δεr,max, is shown to be sensitive to the degree of deformation in the shape of RBCs caused by either impairing the contractile forces acting on the RBCs via a depletion of activated platelets or altering their membrane rigidity. Furthermore, a baseline range for Δεr,max parameter is established using 95 healthy whole blood samples, and the ClotChip is shown to detect weak clot formation, when the platelet count drops below 15,000 per μL. This work highlights the potential of ClotChip as a platform technology for rapid assessment of platelet loss during traumatic hemorrhage at the point-of-injury.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"30 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88801307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956925
J. Bonhomme, M. Oudich, P. A. S. Chavez, L. Olive, D. Beyssen, P. Charette, F. Sarry
In this work, we investigate the evolution of the properties of a pillar phononic crystal in order to simplify the fabrication process by reducing the thickness of the layer. The structure is constructed by stacking alternate layers of SiO2 and Tungsten. The reduction of the dimensions moves the band gap to the upper frequencies and it is possible to adjust its aperture with the ratio comparing the radius of the cylinder with the layers’ thickness. On a substrate, the pillars give resonant modes that can be excited by shear surface waves and gives a sharp peak in the transmission spectrum of the surface wave with a very high quality factor. We also study the theoretical mass sensitivity of the device and we propose a numerical method to determine the theoretical maximum sensitivity that could be obtain by the pillars system. This pillar based metasurface presents a very promising mass sensitivity and a possible way to increase the performance of SAW bio-sensors.
{"title":"Theoretical study of a phononic structure for bio-sensing applications","authors":"J. Bonhomme, M. Oudich, P. A. S. Chavez, L. Olive, D. Beyssen, P. Charette, F. Sarry","doi":"10.1109/SENSORS43011.2019.8956925","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956925","url":null,"abstract":"In this work, we investigate the evolution of the properties of a pillar phononic crystal in order to simplify the fabrication process by reducing the thickness of the layer. The structure is constructed by stacking alternate layers of SiO2 and Tungsten. The reduction of the dimensions moves the band gap to the upper frequencies and it is possible to adjust its aperture with the ratio comparing the radius of the cylinder with the layers’ thickness. On a substrate, the pillars give resonant modes that can be excited by shear surface waves and gives a sharp peak in the transmission spectrum of the surface wave with a very high quality factor. We also study the theoretical mass sensitivity of the device and we propose a numerical method to determine the theoretical maximum sensitivity that could be obtain by the pillars system. This pillar based metasurface presents a very promising mass sensitivity and a possible way to increase the performance of SAW bio-sensors.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"1 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88809127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956782
D. Kocur, T. Porteleky, M. Švecová
In this paper, we will introduce an ultra-wideband (UWB) radar testbed system for short-range localization of static persons in 2-dimensional space for line-of-sight and non-line-of-sight scenarios. The proposed UWB localization system allows localizing multiple static human targets including scenarios with a time-variable number of persons. The measurement rate of the testbed is 1 scan of monitored area per second, and hence the localization system performance can be considered from the application point of view as real-time operating. It is assumed, the introduced UWB radar testbed localization system can be applied as an efficient developed tool of various applications of UWB localization systems of static persons.
{"title":"UWB Radar Testbed System for Localization of Multiple Static Persons","authors":"D. Kocur, T. Porteleky, M. Švecová","doi":"10.1109/SENSORS43011.2019.8956782","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956782","url":null,"abstract":"In this paper, we will introduce an ultra-wideband (UWB) radar testbed system for short-range localization of static persons in 2-dimensional space for line-of-sight and non-line-of-sight scenarios. The proposed UWB localization system allows localizing multiple static human targets including scenarios with a time-variable number of persons. The measurement rate of the testbed is 1 scan of monitored area per second, and hence the localization system performance can be considered from the application point of view as real-time operating. It is assumed, the introduced UWB radar testbed localization system can be applied as an efficient developed tool of various applications of UWB localization systems of static persons.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"13 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88844100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956939
Bhargav Gadhavi, F. Golnaraghi, B. Bahreyni
This paper focuses on the design of a tuning fork microresonator that utilizes its inherent inertial nonlinearities for its operation. Two orthogonal in-plane modes of the microresonator are carefully designed so that the natural frequency ratio between the higher and the lower frequency modes is close to two. When we electrostatically excite the higher frequency mode with the excitation voltage higher than some threshold, the lower frequency mode gets autoparametrically excited due to a nonlinear resonance phenomenon known as two-to-one (2:1) internal resonance. We employed the MEMS Integrated Design for Inertial Sensors (MIDIS), a high vacuum bulk-micromachining process, by Teledyne Dalsa to fabricate the proposed microresonator. We then experimentally demonstrate the 2:1 internal resonance phenomenon occurring in the resonator and highlight its applicability in developing the resonating MEMS based sensors.
{"title":"Design and Characterization of a Tuning Fork Microresonator Based on Nonlinear 2:1 Internal Resonance","authors":"Bhargav Gadhavi, F. Golnaraghi, B. Bahreyni","doi":"10.1109/SENSORS43011.2019.8956939","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956939","url":null,"abstract":"This paper focuses on the design of a tuning fork microresonator that utilizes its inherent inertial nonlinearities for its operation. Two orthogonal in-plane modes of the microresonator are carefully designed so that the natural frequency ratio between the higher and the lower frequency modes is close to two. When we electrostatically excite the higher frequency mode with the excitation voltage higher than some threshold, the lower frequency mode gets autoparametrically excited due to a nonlinear resonance phenomenon known as two-to-one (2:1) internal resonance. We employed the MEMS Integrated Design for Inertial Sensors (MIDIS), a high vacuum bulk-micromachining process, by Teledyne Dalsa to fabricate the proposed microresonator. We then experimentally demonstrate the 2:1 internal resonance phenomenon occurring in the resonator and highlight its applicability in developing the resonating MEMS based sensors.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"9 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88985487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956742
Elliot Yeung, I. AlMohimeed, Y. Ono
Skeletal muscle monitoring provides valuable information about the muscle contractile properties which could be useful in various clinical applications. A wearable ultrasonic sensor (WUS) was constructed using a 110-µm thick polyvinylidene fluoride polymer piezoelectric film to perform a continuous and hands-free muscle monitoring. The lightweight and flexible properties of WUS enables the stable attachment to the skin surface without affecting the tissue motion of interest which results in less motion artifacts. In order to perform reliable estimation of tissue thicknesses by the WUS in an ultrasonic pulse echo mode, the accuracy of selected signal analysis techniques was evaluated using a laser displacement measurement result as a reference. In addition, the effects of moving averaging on the signal-to-noise ratio of the ultrasonic signals and the estimation accuracy of the tissue thickness were investigated. In order to demonstrate the feasibility of the proposed ultrasound technique for in-vivo measurements, continuous monitoring of the lateral head of triceps muscle of a healthy male subject was performed using the WUS. The tissue thickness changes due to the evoked muscle contraction by an electrical muscle stimulation (EMS) were monitored. It was observed that the muscle was unable to fully relax at the EMS frequency at 16 Hz, indicating the tetanic contraction.
{"title":"Estimation of Tissue Thickness Changes Due to Electrical Muscle Stimulation Using Wearable Ultrasonic Sensor in Pulse Echo Mode","authors":"Elliot Yeung, I. AlMohimeed, Y. Ono","doi":"10.1109/SENSORS43011.2019.8956742","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956742","url":null,"abstract":"Skeletal muscle monitoring provides valuable information about the muscle contractile properties which could be useful in various clinical applications. A wearable ultrasonic sensor (WUS) was constructed using a 110-µm thick polyvinylidene fluoride polymer piezoelectric film to perform a continuous and hands-free muscle monitoring. The lightweight and flexible properties of WUS enables the stable attachment to the skin surface without affecting the tissue motion of interest which results in less motion artifacts. In order to perform reliable estimation of tissue thicknesses by the WUS in an ultrasonic pulse echo mode, the accuracy of selected signal analysis techniques was evaluated using a laser displacement measurement result as a reference. In addition, the effects of moving averaging on the signal-to-noise ratio of the ultrasonic signals and the estimation accuracy of the tissue thickness were investigated. In order to demonstrate the feasibility of the proposed ultrasound technique for in-vivo measurements, continuous monitoring of the lateral head of triceps muscle of a healthy male subject was performed using the WUS. The tissue thickness changes due to the evoked muscle contraction by an electrical muscle stimulation (EMS) were monitored. It was observed that the muscle was unable to fully relax at the EMS frequency at 16 Hz, indicating the tetanic contraction.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"73 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86407959","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956853
M. Forssell, G. Fedder, M. Sciullo, Chenchen Mou, Fan Sun, Tyler W. Simpson, Gutian Xiao, L. Fisher, C. Bettinger, C. C. Horn
This paper describes the design of a peripheral nerve cuff electrodes used to stimulate the rat vagus nerve. Multiple electrodes on the circumference of the cuff allow for selective stimulation of the nerve, targeting different functional pathways, including those involved in inflammatory, cardiac, and metabolic diseases. Our approach was verified by stimulating the cervical vagus while recording propagated action potentials to the abdominal vagus nerve. The flexible probe adjusts to the diameter of the nerve, with a robust electrode contact maintained by an adhesive hydrogel.
{"title":"Compliant adhesive cuff electrode for selective stimulation in rat vagus nerve","authors":"M. Forssell, G. Fedder, M. Sciullo, Chenchen Mou, Fan Sun, Tyler W. Simpson, Gutian Xiao, L. Fisher, C. Bettinger, C. C. Horn","doi":"10.1109/SENSORS43011.2019.8956853","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956853","url":null,"abstract":"This paper describes the design of a peripheral nerve cuff electrodes used to stimulate the rat vagus nerve. Multiple electrodes on the circumference of the cuff allow for selective stimulation of the nerve, targeting different functional pathways, including those involved in inflammatory, cardiac, and metabolic diseases. Our approach was verified by stimulating the cervical vagus while recording propagated action potentials to the abdominal vagus nerve. The flexible probe adjusts to the diameter of the nerve, with a robust electrode contact maintained by an adhesive hydrogel.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"31 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81259535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/SENSORS43011.2019.8956721
M. Belmonte, A. Casson, Niels Peek
Long-Short Term Memory models (LSTMs) are data-driven routines that classify Human Activity Recognition (HAR) with minimum human input. The price to pay for analysing large sequences of on-body sensor measurements with LSTMs are high processing power and battery requirements. In this paper, we recognize that sensor data packets have differing information value to classify HAR and propose to quantify it with cross entropy (CrossEn), Kullback Leibler (KL) divergence and sample entropy (SampEn). Both, CrossEn and SampEn have the potential to guide dropping redundant data packets without compromising HAR. However, we do not find substantial improvements in dropping rates when downsampling by CrossEn and SampEn over computationally cheaper random and uniform alternatives. Our results show that the KL divergence, evaluated at training time is equivalent to the classification accuracy criteria that involves a testing set. The computational requirements to compute the KL in real-time could well guide sensor node design to downsample wearable measurements near the user.
{"title":"Downsampling wearable sensor data packets by measuring their information value","authors":"M. Belmonte, A. Casson, Niels Peek","doi":"10.1109/SENSORS43011.2019.8956721","DOIUrl":"https://doi.org/10.1109/SENSORS43011.2019.8956721","url":null,"abstract":"Long-Short Term Memory models (LSTMs) are data-driven routines that classify Human Activity Recognition (HAR) with minimum human input. The price to pay for analysing large sequences of on-body sensor measurements with LSTMs are high processing power and battery requirements. In this paper, we recognize that sensor data packets have differing information value to classify HAR and propose to quantify it with cross entropy (CrossEn), Kullback Leibler (KL) divergence and sample entropy (SampEn). Both, CrossEn and SampEn have the potential to guide dropping redundant data packets without compromising HAR. However, we do not find substantial improvements in dropping rates when downsampling by CrossEn and SampEn over computationally cheaper random and uniform alternatives. Our results show that the KL divergence, evaluated at training time is equivalent to the classification accuracy criteria that involves a testing set. The computational requirements to compute the KL in real-time could well guide sensor node design to downsample wearable measurements near the user.","PeriodicalId":6710,"journal":{"name":"2019 IEEE SENSORS","volume":"17 1","pages":"1-4"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85022717","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：