Pub Date : 2016-01-18DOI: 10.1109/TBCAS.2015.2495341
H. Toreyin, P. Bhatti
A low-power ASIC signal processor for a vestibular prosthesis (VP) is reported. Fabricated with TI 0.35 μm CMOS technology and designed to interface with implanted inertial sensors, the digitally assisted analog signal processor operates extensively in the CMOS subthreshold region. During its operation the ASIC encodes head motion signals captured by the inertial sensors as electrical pulses ultimately targeted for in-vivo stimulation of vestibular nerve fibers. To achieve this, the ASIC implements a coordinate system transformation to correct for misalignment between natural sensors and implanted inertial sensors. It also mimics the frequency response characteristics and frequency encoding mappings of angular and linear head motions observed at the peripheral sense organs, semicircular canals and otolith. Overall the design occupies an area of 6.22 mm 2 and consumes 1.24 mW when supplied with ± 1.6 V.
报道了一种用于前庭假体(VP)的低功耗ASIC信号处理器。该数字辅助模拟信号处理器采用TI 0.35 μm CMOS技术制造，设计用于与植入惯性传感器接口，在CMOS亚阈值区域广泛工作。在操作过程中，ASIC将惯性传感器捕获的头部运动信号编码为电脉冲，最终用于在体内刺激前庭神经纤维。为了实现这一点，ASIC实现了坐标系转换，以纠正自然传感器和植入惯性传感器之间的不对准。它还模拟了外围感觉器官、半规管和耳石观察到的角状和线性头部运动的频率响应特性和频率编码映射。总体而言，该设计占地6.22 mm 2，在±1.6 V供电时消耗1.24 mW。
{"title":"A Low-Power ASIC Signal Processor for a Vestibular Prosthesis","authors":"H. Toreyin, P. Bhatti","doi":"10.1109/TBCAS.2015.2495341","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2495341","url":null,"abstract":"A low-power ASIC signal processor for a vestibular prosthesis (VP) is reported. Fabricated with TI 0.35 μm CMOS technology and designed to interface with implanted inertial sensors, the digitally assisted analog signal processor operates extensively in the CMOS subthreshold region. During its operation the ASIC encodes head motion signals captured by the inertial sensors as electrical pulses ultimately targeted for in-vivo stimulation of vestibular nerve fibers. To achieve this, the ASIC implements a coordinate system transformation to correct for misalignment between natural sensors and implanted inertial sensors. It also mimics the frequency response characteristics and frequency encoding mappings of angular and linear head motions observed at the peripheral sense organs, semicircular canals and otolith. Overall the design occupies an area of 6.22 mm 2 and consumes 1.24 mW when supplied with ± 1.6 V.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"10 1","pages":"768-778"},"PeriodicalIF":5.1,"publicationDate":"2016-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2495341","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-01-08DOI: 10.1109/TBCAS.2015.2490225
G. Massicotte, S. Carrara, G. Micheli, M. Sawan
In vivo multi-target and selective concentration monitoring of neurotransmitters can help to unravel the brain chemical complex signaling interplay. This paper presents a dedicated integrated potentiostat transducer circuit and its selective electrode interface. A custom 2-electrode time-based potentiostat circuit was fabricated with 0.13 μm CMOS technology and provides a wide dynamic input current range of 20 pA to 600 nA with 56 μW, for a minimum sampling frequency of 1.25 kHz. A multi-working electrode chip is functionalized with carbon nanotubes (CNT)-based chemical coatings that offer high sensitivity and selectivity towards electroactive dopamine and non-electroactive glutamate. The prototype was experimentally tested with different concentrations levels of both neurotransmitter types, and results were similar to measurements with a commercially available potentiostat. This paper validates the functionality of the proposed biosensor, and demonstrates its potential for the selective detection of a large number of neurochemicals.
体内多靶点和选择性的神经递质浓度监测有助于揭示大脑化学复合物信号的相互作用。本文介绍了一种专用的集成恒电位器电路及其选择电极接口。采用0.13 μm CMOS工艺制作了一种定制的2电极时基恒电位器电路，其动态输入电流范围为20 pA ~ 600 nA，功率为56 μW，最小采样频率为1.25 kHz。采用基于碳纳米管(CNT)的化学涂层对多工作电极芯片进行功能化，该涂层对电活性多巴胺和非电活性谷氨酸具有高灵敏度和选择性。该原型在两种神经递质类型的不同浓度水平下进行了实验测试，结果与市售的恒电位器测量结果相似。本文验证了所提出的生物传感器的功能，并展示了其选择性检测大量神经化学物质的潜力。
{"title":"A CMOS Amperometric System for Multi-Neurotransmitter Detection","authors":"G. Massicotte, S. Carrara, G. Micheli, M. Sawan","doi":"10.1109/TBCAS.2015.2490225","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2490225","url":null,"abstract":"In vivo multi-target and selective concentration monitoring of neurotransmitters can help to unravel the brain chemical complex signaling interplay. This paper presents a dedicated integrated potentiostat transducer circuit and its selective electrode interface. A custom 2-electrode time-based potentiostat circuit was fabricated with 0.13 μm CMOS technology and provides a wide dynamic input current range of 20 pA to 600 nA with 56 μW, for a minimum sampling frequency of 1.25 kHz. A multi-working electrode chip is functionalized with carbon nanotubes (CNT)-based chemical coatings that offer high sensitivity and selectivity towards electroactive dopamine and non-electroactive glutamate. The prototype was experimentally tested with different concentrations levels of both neurotransmitter types, and results were similar to measurements with a commercially available potentiostat. This paper validates the functionality of the proposed biosensor, and demonstrates its potential for the selective detection of a large number of neurochemicals.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"10 1","pages":"731-741"},"PeriodicalIF":5.1,"publicationDate":"2016-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2490225","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62964957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-01-06DOI: 10.1109/TBCAS.2015.2487603
B. C. Cheah, A. Macdonald, Christopher Martin, A. Streklas, Gordon Campbell, M. Al-Rawhani, B. Németh, J. Grant, M. Barrett, D. Cumming
We have created a novel chip-based diagnostic tools based upon quantification of metabolites using enzymes specific for their chemical conversion. Using this device we show for the first time that a solid-state circuit can be used to measure enzyme kinetics and calculate the Michaelis-Menten constant. Substrate concentration dependency of enzyme reaction rates is central to this aim. Ion-sensitive field effect transistors (ISFET) are excellent transducers for biosensing applications that are reliant upon enzyme assays, especially since they can be fabricated using mainstream microelectronics technology to ensure low unit cost, mass-manufacture, scaling to make many sensors and straightforward miniaturisation for use in point-of-care devices. Here, we describe an integrated ISFET array comprising 216 sensors. The device was fabricated with a complementary metal oxide semiconductor (CMOS) process. Unlike traditional CMOS ISFET sensors that use the Si3N4 passivation of the foundry for ion detection, the device reported here was processed with a layer of Ta2O5 that increased the detection sensitivity to 45 mV/pH unit at the sensor readout. The drift was reduced to 0.8 mV/hour with a linear pH response between pH 2-12. A high-speed instrumentation system capable of acquiring nearly 500 fps was developed to stream out the data. The device was then used to measure glucose concentration through the activity of hexokinase in the range of 0.05 mM-231 mM, encompassing glucose's physiological range in blood. Localised and temporal enzyme kinetics of hexokinase was studied in detail. These results present a roadmap towards a viable personal metabolome machine.
{"title":"An Integrated Circuit for Chip-Based Analysis of Enzyme Kinetics and Metabolite Quantification","authors":"B. C. Cheah, A. Macdonald, Christopher Martin, A. Streklas, Gordon Campbell, M. Al-Rawhani, B. Németh, J. Grant, M. Barrett, D. Cumming","doi":"10.1109/TBCAS.2015.2487603","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2487603","url":null,"abstract":"We have created a novel chip-based diagnostic tools based upon quantification of metabolites using enzymes specific for their chemical conversion. Using this device we show for the first time that a solid-state circuit can be used to measure enzyme kinetics and calculate the Michaelis-Menten constant. Substrate concentration dependency of enzyme reaction rates is central to this aim. Ion-sensitive field effect transistors (ISFET) are excellent transducers for biosensing applications that are reliant upon enzyme assays, especially since they can be fabricated using mainstream microelectronics technology to ensure low unit cost, mass-manufacture, scaling to make many sensors and straightforward miniaturisation for use in point-of-care devices. Here, we describe an integrated ISFET array comprising 216 sensors. The device was fabricated with a complementary metal oxide semiconductor (CMOS) process. Unlike traditional CMOS ISFET sensors that use the Si3N4 passivation of the foundry for ion detection, the device reported here was processed with a layer of Ta2O5 that increased the detection sensitivity to 45 mV/pH unit at the sensor readout. The drift was reduced to 0.8 mV/hour with a linear pH response between pH 2-12. A high-speed instrumentation system capable of acquiring nearly 500 fps was developed to stream out the data. The device was then used to measure glucose concentration through the activity of hexokinase in the range of 0.05 mM-231 mM, encompassing glucose's physiological range in blood. Localised and temporal enzyme kinetics of hexokinase was studied in detail. These results present a roadmap towards a viable personal metabolome machine.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"10 1","pages":"721-730"},"PeriodicalIF":5.1,"publicationDate":"2016-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2487603","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62964927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2016-01-04DOI: 10.1109/TBCAS.2015.2501359
Mohsen Shokouhian, R. Morling, I. Kale
Ambient light and optical interference can severely affect the performance of pulse oximeters. The deployment of a robust modulation technique to drive the pulse oximeter LEDs can reduce these unwanted effects and increases the resilient of the pulse oximeter against artificial ambient light. The time division modulation technique used in conventional pulse oximeters can not remove the effect of modulated light coming from surrounding environment and this may cause huge measurement error in pulse oximeter readings. This paper presents a novel cross-coupled sigma delta modulator which ensures that measurement accuracy will be more robust in comparison with conventional fixed-frequency oximeter modulation technique especially in the presence of pulsed artificial ambient light. Moreover, this novel modulator gives an extra control over the pulse oximeter power consumption leading to improved power management.
{"title":"Interference Resilient Sigma Delta-Based Pulse Oximeter","authors":"Mohsen Shokouhian, R. Morling, I. Kale","doi":"10.1109/TBCAS.2015.2501359","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2501359","url":null,"abstract":"Ambient light and optical interference can severely affect the performance of pulse oximeters. The deployment of a robust modulation technique to drive the pulse oximeter LEDs can reduce these unwanted effects and increases the resilient of the pulse oximeter against artificial ambient light. The time division modulation technique used in conventional pulse oximeters can not remove the effect of modulated light coming from surrounding environment and this may cause huge measurement error in pulse oximeter readings. This paper presents a novel cross-coupled sigma delta modulator which ensures that measurement accuracy will be more robust in comparison with conventional fixed-frequency oximeter modulation technique especially in the presence of pulsed artificial ambient light. Moreover, this novel modulator gives an extra control over the pulse oximeter power consumption leading to improved power management.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"10 1","pages":"623-631"},"PeriodicalIF":5.1,"publicationDate":"2016-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2501359","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-29DOI: 10.1109/TBCAS.2015.2500911
Jin-Hong Ahn, Sang-Hoon Hong, Youngjune Park
This paper presents a double-side CMOS-carbon nanotube (CNT) sensor array for simple bare-die measurements in a medical environment based on a 0.35 μm standard CMOS process. This scheme allows robust measurements due to its inherent back-side rectifying diodes with a high latch-up resistance. In particular, instead of using pads, only two contact metal structures: a wide ring structure around the sensor area on the front side and a plate structure at the backside are used for both power and single I/O line. The back-side rectification is made possible by creating VDD and VSS through the back-side and front-side, respectively. The single I/O line is conditioned such that it doubles as either the power source or the ground, depending on whether the chip is face down or face up. A modified universal asynchronous receiver/transmitter (UART) serial communication scheme with pulse based I/O signal transmission is developed to reduce the power degradation during the signaling intervals. In addition, communication errors and I/O power dissipation for the receiver path are minimized by using level sensitive switch control and double sampling difference amplifier. In order to implement these special functions, a controller chip with a special I/O protocol is designed. Using this controller chip, issuing commands and receiving data can both be performed on a single line and the results are flexibly measured through either the backside or the front side of the chip contacts. As a result, a stable operation of under 150 mW maximum power at 2 MHz data rate can be achieved. The double-side chips with 32 × 32 and 64 × 64 sensor arrays occupy areas of 1.9×2.3 mm2 and 3.7×3.9 mm2, respectively.
{"title":"A Double-Side CMOS-CNT Biosensor Array With Padless Structure for Simple Bare-Die Measurements in a Medical Environment","authors":"Jin-Hong Ahn, Sang-Hoon Hong, Youngjune Park","doi":"10.1109/TBCAS.2015.2500911","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2500911","url":null,"abstract":"This paper presents a double-side CMOS-carbon nanotube (CNT) sensor array for simple bare-die measurements in a medical environment based on a 0.35 μm standard CMOS process. This scheme allows robust measurements due to its inherent back-side rectifying diodes with a high latch-up resistance. In particular, instead of using pads, only two contact metal structures: a wide ring structure around the sensor area on the front side and a plate structure at the backside are used for both power and single I/O line. The back-side rectification is made possible by creating VDD and VSS through the back-side and front-side, respectively. The single I/O line is conditioned such that it doubles as either the power source or the ground, depending on whether the chip is face down or face up. A modified universal asynchronous receiver/transmitter (UART) serial communication scheme with pulse based I/O signal transmission is developed to reduce the power degradation during the signaling intervals. In addition, communication errors and I/O power dissipation for the receiver path are minimized by using level sensitive switch control and double sampling difference amplifier. In order to implement these special functions, a controller chip with a special I/O protocol is designed. Using this controller chip, issuing commands and receiving data can both be performed on a single line and the results are flexibly measured through either the backside or the front side of the chip contacts. As a result, a stable operation of under 150 mW maximum power at 2 MHz data rate can be achieved. The double-side chips with 32 × 32 and 64 × 64 sensor arrays occupy areas of 1.9×2.3 mm2 and 3.7×3.9 mm2, respectively.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"815-824"},"PeriodicalIF":5.1,"publicationDate":"2015-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2500911","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-28DOI: 10.1109/TBCAS.2015.2498643
Abhishek Roy, Alicia Klinefelter, Farah B. Yahya, Xing Chen, Luis Gonzalez-Guerrero, Christopher J. Lukas, Divya Akella, James Boley, Kyle Craig, M. Faisal, Seunghyun Oh, N. Roberts, Y. Shakhsheer, A. Shrivastava, D. Vasudevan, D. Wentzloff, B. Calhoun
This paper presents a batteryless system-on-chip (SoC) that operates off energy harvested from indoor solar cells and/or thermoelectric generators (TEGs) on the body. Fabricated in a commercial 0.13 μW process, this SoC sensing platform consists of an integrated energy harvesting and power management unit (EH-PMU) with maximum power point tracking, multiple sensing modalities, programmable core and a low power microcontroller with several hardware accelerators to enable energy-efficient digital signal processing, ultra-low-power (ULP) asymmetric radios for wireless transmission, and a 100 nW wake-up radio. The EH-PMU achieves a peak end-to-end efficiency of 75% delivering power to a 100 μA load. In an example motion detection application, the SoC reads data from an accelerometer through SPI, processes it, and sends it over the radio. The SPI and digital processing consume only 2.27 μW, while the integrated radio consumes 4.18 μW when transmitting at 187.5 kbps for a total of 6.45 μW.
{"title":"A 6.45 $mu{rm W}$ Self-Powered SoC With Integrated Energy-Harvesting Power Management and ULP Asymmetric Radios for Portable Biomedical Systems","authors":"Abhishek Roy, Alicia Klinefelter, Farah B. Yahya, Xing Chen, Luis Gonzalez-Guerrero, Christopher J. Lukas, Divya Akella, James Boley, Kyle Craig, M. Faisal, Seunghyun Oh, N. Roberts, Y. Shakhsheer, A. Shrivastava, D. Vasudevan, D. Wentzloff, B. Calhoun","doi":"10.1109/TBCAS.2015.2498643","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2498643","url":null,"abstract":"This paper presents a batteryless system-on-chip (SoC) that operates off energy harvested from indoor solar cells and/or thermoelectric generators (TEGs) on the body. Fabricated in a commercial 0.13 μW process, this SoC sensing platform consists of an integrated energy harvesting and power management unit (EH-PMU) with maximum power point tracking, multiple sensing modalities, programmable core and a low power microcontroller with several hardware accelerators to enable energy-efficient digital signal processing, ultra-low-power (ULP) asymmetric radios for wireless transmission, and a 100 nW wake-up radio. The EH-PMU achieves a peak end-to-end efficiency of 75% delivering power to a 100 μA load. In an example motion detection application, the SoC reads data from an accelerometer through SPI, processes it, and sends it over the radio. The SPI and digital processing consume only 2.27 μW, while the integrated radio consumes 4.18 μW when transmitting at 187.5 kbps for a total of 6.45 μW.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"862-874"},"PeriodicalIF":5.1,"publicationDate":"2015-12-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2498643","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62964930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-28DOI: 10.1109/TBCAS.2015.2500101
Zhuo Wang, Jintao Zhang, N. Verma
In wearable and implantable medical-sensor applications, low-energy classification systems are of importance for deriving high-quality inferences locally within the device. Given that sensor instrumentation is typically followed by A-D conversion, this paper presents a system implementation wherein the majority of the computations required for classification are implemented within the ADC. To achieve this, first an algorithmic formulation is presented that combines linear feature extraction and classification into a single matrix transformation. Second, a matrix-multiplying ADC (MMADC) is presented that enables multiplication between an analog input sample and a digital multiplier, with negligible additional energy beyond that required for A-D conversion. Two systems mapped to the MMADC are demonstrated: (1) an ECG-based cardiac arrhythmia detector; and (2) an image-pixel-based facial gender detector. The RMS error over all multiplication performed, normalized to the RMS of ideal multiplication results is 0.018. Further, compared to idealized versions of conventional systems, the energy savings obtained are estimated to be 13× and 29×, respectively, while achieving similar level of performance.
{"title":"Realizing Low-Energy Classification Systems by Implementing Matrix Multiplication Directly Within an ADC","authors":"Zhuo Wang, Jintao Zhang, N. Verma","doi":"10.1109/TBCAS.2015.2500101","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2500101","url":null,"abstract":"In wearable and implantable medical-sensor applications, low-energy classification systems are of importance for deriving high-quality inferences locally within the device. Given that sensor instrumentation is typically followed by A-D conversion, this paper presents a system implementation wherein the majority of the computations required for classification are implemented within the ADC. To achieve this, first an algorithmic formulation is presented that combines linear feature extraction and classification into a single matrix transformation. Second, a matrix-multiplying ADC (MMADC) is presented that enables multiplication between an analog input sample and a digital multiplier, with negligible additional energy beyond that required for A-D conversion. Two systems mapped to the MMADC are demonstrated: (1) an ECG-based cardiac arrhythmia detector; and (2) an image-pixel-based facial gender detector. The RMS error over all multiplication performed, normalized to the RMS of ideal multiplication results is 0.018. Further, compared to idealized versions of conventional systems, the energy savings obtained are estimated to be 13× and 29×, respectively, while achieving similar level of performance.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"825-837"},"PeriodicalIF":5.1,"publicationDate":"2015-12-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2500101","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965000","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-01DOI: 10.1109/TBCAS.2015.2504563
Seongwook Park, Junyoung Park, Kyeongryeol Bong, Dongjoo Shin, Jinmook Lee, Sungpill Choi, H. Yoo
Deep Learning algorithm is widely used for various pattern recognition applications such as text recognition, object recognition and action recognition because of its best-in-class recognition accuracy compared to hand-crafted algorithm and shallow learning based algorithms. Long learning time caused by its complex structure, however, limits its usage only in high-cost servers or many-core GPU platforms so far. On the other hand, the demand on customized pattern recognition within personal devices will grow gradually as more deep learning applications will be developed. This paper presents a SoC implementation to enable deep learning applications to run with low cost platforms such as mobile or portable devices. Different from conventional works which have adopted massively-parallel architecture, this work adopts task-flexible architecture and exploits multiple parallelism to cover complex functions of convolutional deep belief network which is one of popular deep learning/inference algorithms. In this paper, we implement the most energy-efficient deep learning and inference processor for wearable system. The implemented 2.5 mm ×4.0 mm deep learning/inference processor is fabricated using 65 nm 8-metal CMOS technology for a battery-powered platform with real-time deep inference and deep learning operation. It consumes 185 mW average power, and 213.1 mW peak power at 200 MHz operating frequency and 1.2 V supply voltage. It achieves 411.3 GOPS peak performance and 1.93 TOPS/W energy efficiency, which is 2.07× higher than the state-of-the-art.
深度学习算法被广泛用于各种模式识别应用，如文本识别、对象识别和动作识别，因为与手工算法和基于浅学习的算法相比，深度学习算法具有同类最佳的识别精度。但由于其结构复杂，学习时间长，目前仅局限于高成本服务器或多核GPU平台。另一方面，随着更多深度学习应用的开发，个人设备对定制模式识别的需求将逐渐增长。本文提出了一个SoC实现，使深度学习应用程序能够在低成本平台(如移动或便携式设备)上运行。与传统的大规模并行架构不同，本文采用了任务柔性架构，利用多重并行性覆盖了卷积深度信念网络的复杂功能，卷积深度信念网络是目前流行的深度学习/推理算法之一。在本文中，我们为可穿戴系统实现了最节能的深度学习和推理处理器。实现的2.5 mm ×4.0 mm深度学习/推理处理器采用65 nm 8金属CMOS技术制造，用于具有实时深度推理和深度学习操作的电池供电平台。在200mhz工作频率和1.2 V电源电压下，平均功耗为185mw，峰值功耗为213.1 mW。峰值性能达到411.3 GOPS，能效达到1.93 TOPS/W，比目前先进水平提高2.07倍。
{"title":"An Energy-Efficient and Scalable Deep Learning/Inference Processor With Tetra-Parallel MIMD Architecture for Big Data Applications","authors":"Seongwook Park, Junyoung Park, Kyeongryeol Bong, Dongjoo Shin, Jinmook Lee, Sungpill Choi, H. Yoo","doi":"10.1109/TBCAS.2015.2504563","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2504563","url":null,"abstract":"Deep Learning algorithm is widely used for various pattern recognition applications such as text recognition, object recognition and action recognition because of its best-in-class recognition accuracy compared to hand-crafted algorithm and shallow learning based algorithms. Long learning time caused by its complex structure, however, limits its usage only in high-cost servers or many-core GPU platforms so far. On the other hand, the demand on customized pattern recognition within personal devices will grow gradually as more deep learning applications will be developed. This paper presents a SoC implementation to enable deep learning applications to run with low cost platforms such as mobile or portable devices. Different from conventional works which have adopted massively-parallel architecture, this work adopts task-flexible architecture and exploits multiple parallelism to cover complex functions of convolutional deep belief network which is one of popular deep learning/inference algorithms. In this paper, we implement the most energy-efficient deep learning and inference processor for wearable system. The implemented 2.5 mm ×4.0 mm deep learning/inference processor is fabricated using 65 nm 8-metal CMOS technology for a battery-powered platform with real-time deep inference and deep learning operation. It consumes 185 mW average power, and 213.1 mW peak power at 200 MHz operating frequency and 1.2 V supply voltage. It achieves 411.3 GOPS peak performance and 1.93 TOPS/W energy efficiency, which is 2.07× higher than the state-of-the-art.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"838-848"},"PeriodicalIF":5.1,"publicationDate":"2015-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2504563","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62966130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-01DOI: 10.1109/TBCAS.2015.2507618
Po-Hung Kuo, Jui-Chang Kuo, Hsiao-Ting Hsueh, J. Hsieh, Yi-Chun Huang, Tao Wang, Yen-Hung Lin, Chih-Ting Lin, Yao-Joe Yang, Shey-Shi Lu
A micro-controller unit (MCU) assisted immunoassay lab-on-a-chip is realized in 0.35 μm CMOS technology. The MCU automatically controls the detection procedure including blood filtration through a nonporous aluminum oxide membrane, bimolecular conjugation with antibodies attached to magnetic beads, electrolytic pumping, magnetic flushing and threshold detection based on Hall sensor array readout analysis. To verify the function of this chip, in-vitro Tumor necrosis factor- α (TNF- α) and N-terminal pro-brain natriuretic peptide (NT-proBNP) tests are performed by this 9 mm 2-sized single chip. The cost, efficiency and portability are considerably improved compared to the prior art.
{"title":"A Smart CMOS Assay SoC for Rapid Blood Screening Test of Risk Prediction","authors":"Po-Hung Kuo, Jui-Chang Kuo, Hsiao-Ting Hsueh, J. Hsieh, Yi-Chun Huang, Tao Wang, Yen-Hung Lin, Chih-Ting Lin, Yao-Joe Yang, Shey-Shi Lu","doi":"10.1109/TBCAS.2015.2507618","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2507618","url":null,"abstract":"A micro-controller unit (MCU) assisted immunoassay lab-on-a-chip is realized in 0.35 μm CMOS technology. The MCU automatically controls the detection procedure including blood filtration through a nonporous aluminum oxide membrane, bimolecular conjugation with antibodies attached to magnetic beads, electrolytic pumping, magnetic flushing and threshold detection based on Hall sensor array readout analysis. To verify the function of this chip, in-vitro Tumor necrosis factor- α (TNF- α) and N-terminal pro-brain natriuretic peptide (NT-proBNP) tests are performed by this 9 mm 2-sized single chip. The cost, efficiency and portability are considerably improved compared to the prior art.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"790-800"},"PeriodicalIF":5.1,"publicationDate":"2015-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2507618","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965698","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2015-12-01DOI: 10.1109/TBCAS.2015.2501320
A. Donida, G. Dato, Paolo Cunzolo, M. Sala, Filippo Piffaretti, P. Orsatti, D. Barrettino
This paper presents a new system to measure the Intraocular Pressure (IOP) with very high accuracy (0.036 mbar) used for monitoring glaucoma. The system not only monitors the daily variation of the IOP (circadian IOP), but also allows to perform an spectral analysis of the pressure signal generated by the heartbeat (cardiac IOP). The system comprises a piezoresistive pressure sensor, an application-specific integrated circuit (ASIC) to read out the sensor data and an external reader installed on customized glasses. The ASIC readout electronics combines chopping modulation with correlated double sampling (CDS) in order to eliminate both the amplifier offset and the chopper ripple at the sampling frequency. In addition, programmable current sources are used to compensate for the atmospheric pressure ( 800-1200 mbar ) and the circadian component ( ±7 mbar) thus allowing to read out the very weak cardiac signals ( ±1.6 mbar) with a maximum accuracy of 0.036 mbar.
{"title":"A Circadian and Cardiac Intraocular Pressure Sensor for Smart Implantable Lens","authors":"A. Donida, G. Dato, Paolo Cunzolo, M. Sala, Filippo Piffaretti, P. Orsatti, D. Barrettino","doi":"10.1109/TBCAS.2015.2501320","DOIUrl":"https://doi.org/10.1109/TBCAS.2015.2501320","url":null,"abstract":"This paper presents a new system to measure the Intraocular Pressure (IOP) with very high accuracy (0.036 mbar) used for monitoring glaucoma. The system not only monitors the daily variation of the IOP (circadian IOP), but also allows to perform an spectral analysis of the pressure signal generated by the heartbeat (cardiac IOP). The system comprises a piezoresistive pressure sensor, an application-specific integrated circuit (ASIC) to read out the sensor data and an external reader installed on customized glasses. The ASIC readout electronics combines chopping modulation with correlated double sampling (CDS) in order to eliminate both the amplifier offset and the chopper ripple at the sampling frequency. In addition, programmable current sources are used to compensate for the atmospheric pressure ( 800-1200 mbar ) and the circadian component ( ±7 mbar) thus allowing to read out the very weak cardiac signals ( ±1.6 mbar) with a maximum accuracy of 0.036 mbar.","PeriodicalId":13151,"journal":{"name":"IEEE Transactions on Biomedical Circuits and Systems","volume":"9 1","pages":"777-789"},"PeriodicalIF":5.1,"publicationDate":"2015-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/TBCAS.2015.2501320","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62965192","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","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：