KTH Information and Communication Technology Hybrid Integration of Active Bio-signal Cable with Intelligent Electrode. Steps toward Wearable Pervasive-Healthcare Applications Doctoral Thesis By GENG YANG Electronic and Computer Systems School of Information and Communication Technology Stockholm, Sweden 2012 TRITA-ICT/ECS AVH 13:04 ISSN 1653-6363 ISRN KTH/ICT/ECS/AVH-13/04-SE ISBN 978-91-7501-670-2 Royal Institute of Technology School of Information and Communication Technology Department of Electronic Systems Forum 120. SE-164 40 Kista, SWEDEN Thesis submitted to the School of Information and Communication Technology (ICT), Department of Electronic Systems (ES), Royal Institute of Technology (KTH), to be public defended on 2013-04-16 at 13:00 in Sal E, Forum, Kista, Stockholm, Sweden. Copyright © Geng Yang, 2012 Tryck: Universitetsservice US AB Abstract Personalized and pervasive healthcare help seamlessly integrate healthcare and wellness into people’s daily life, independent of time and space. With the developments in biomedical sensing technologies nowadays, silicon based integrated circuits have shown great advantages in terms of tiny physical size, and low power consumption. As a result, they have been found in many advanced medical applications. In the meanwhile, printed electronics is considered as a promising approach enabling cost-effective manufacturing of thin, flexible, and light-weight devices. A hybrid integration of integrated circuits and printed electronics provides a promising solution for the future wearable healthcare devices. This thesis first reviews the current approaches for bio-electric signal sensing and the state-of-the-art designs for biomedical circuit and systems. In the second part, the idea of Intelligent Electrode and Active Cable for wearable ECG monitoring systems is proposed. Based on this concept, we design and fabricate two customized IC chips to provide a single cable solution for long-term healthcare monitoring. The first chip is a digital ASIC with a serial communication protocol implemented on chip to support data and command packets transmission between different ASIC chips. Also, it has on-chip memory to buffer the digital bio-signal. An Intelligent Electrode is formed by embedding the ASIC chip into the conductive electrode. With the on-chip integrated communication protocol, a wired sensor network can be established enabling the single cable solution. The ASIC’s controlling logic is capable of making dynamic network management, thus endows the electrode with local intelligence. The second chip is a fully integrated mixed-signal SoC. In addition to the digital controller implemented and verified in the first chip, another 2 key modules are integrated: a tunable analog front end circuits, and a 6-input SAR ADC. The second chip works as a networked SoC sensor. The command-based network management is verified through functional tests using the fabricated SoCs. With the programmable analog front end circuits, the SoC sensor can be configured to detect a variety of bio-electric signals. EOG, EMG, ECG, and EEG signals are successfully recorded through in-vivo tests. iii This research also explores the potential of using high accurate inkjet printing technology as an inexpensive integration method and enabling technology to design and fabricate bio-sensing devices. Performance evaluation of printed electrodes and interconnections on flexible substrates is made to examine the feasibility of applying them in the fabrication of Bio-Patch. The reliability of the inkjet printed sliver traces is evaluated via static bending tests. The measurement results prove that the printed silver lines can offer a reliable interconnection. In-vivo test results show that the quality of ECG signal sensed by the printed electrodes is comparable with the one gained by commercial electrodes. Finally, two Bio-Patch prototypes are presented: one is based on photo paper substrate, the other on polyimide substrate. These two prototypes are implemented by heterogeneous integration of the silicon based SoC sensor with cost-effective printed electronics onto the flexible substrates. The measurement results indicate the SoC operates smoothly with the printed electronics. Clean ECG signal is successfully recorded from both of the implemented Bio-Patch prototypes. This versatile SoC sensor can be used in various applications according to specific requirements. And this heterogeneous system combining high-level integrated SoC technology and inkjet printing technique provides a promising solution for future personalized and pervasive healthcare applications. Keywords: ASIC, SoC, bio-medical electronics, bio-sensor, Bio-Patch, wearable healthcare system, pervasive healthcare, CMOS, SAR-ADC, Intelligent Electrode, Active Cable, multi-parameter bio-signal sensor. iv Contents Abstract iii Contents v List of Papers viii List of Acronyms xiii Acknowledgements xv 1. Introduction ..................................................................................................................... 1 1.1 Motivation and Pervasive Healthcare .......................................................................1 1.2 Work Performed ........................................................................................................3 1.3 Author’s Contributions..............................................................................................5 1.4 Thesis Outline ...........................................................................................................7 1.5 Beyond the Scope of this Work.................................................................................8 2. Bio-electrical Signal Sensing and Current Approaches................................................ 9 2.1 Introduction ...............................................................................................................9 2.2 Principles of Bio-potential Measurements ..............................................................10 2.3 Current Approaches for Bio-potential Sensing .......................................................13 2.3.1 Electrode Technology.................................................................................... 13 2.3.2 Latest State-of-the-art Biomedical Circuits and Systems ............................. 15 3. SoC Design, Implementation and in-vivo Tests ........................................................... 17 3.1 SoC Architecture and Implementation ....................................................................17 3.1.1 Digital Core................................................................................................... 18 3.1.2 Bio-Electric SoC ........................................................................................... 21 iii 3.2 SoC in-vivo Tests .................................................................................................... 27 3.3 Conclusion .............................................................................................................. 32 4. System Integration and Bio-Patch Implementation ....................................................33 4.1 Printed Electrode..................................................................................................... 33 4.2 Paper based Bio-Patch ............................................................................................ 37 4.3 Static Bending Tests................................................................................................ 39 4.4 Polyimide (PI) based Bio-Patch.............................................................................. 42 4.5 Conclusion .............................................................................................................. 45 5. Summary and Future Works .........................................................................................47 5.1 Summary ................................................................................................................. 47 5.2 Future Works........................................................................................................... 48 Bibliography ........................................................................................................................51 iv List of Papers The publications produced from the research activities performed in this PhD dissertation are listed below. All the conference papers and the journal articles are peer-reviewed. Among the publications, 7 journal articles and conferences papers have been selected to be appended at the end of this thesis. A brief summary of each of them is presented in the following lines. List of Papers included in the thesis: Journal Papers: Paper I: G. Yang, L. Xie, M. Mäntysalo, J. Chen, T. Hannu, and L.-R. Zheng “Bio-Patch Design and Implementation Based on a Low-Power System-on-Chip and Paper-based Inkjet Printing Technology” IEEE Transactions on Information Technology in Biomedicine (T-ITB), vol.16, no.6, pp.1043-1050, Nov. 2012. Paper II: G. Yang, J. Chen, L. Xie, J. Mao, T. Hannu, and L.-R. Zheng “A Hybrid Low Power Bio-Patch for Body Surface Potential Measurement” IEEE Journal of Biomedical and Health Informatics (J-BHI), 2013 (accepted). Conference Papers: Paper III: G. Yang, J. Chen, F. Jonsson, T. Hannu, and L- R. Zheng, “A multi-parameter bio-electric ASIC sensor with integrated 2-wire data transmission protocol for wearable healthcare system,” Design, Automation & Test in Europe (DATE 2012) pp. 443-448. Mar. 2012. Paper IV: L. Xie, G. Yang, M. Mäntysalo, F. Jonsson, and L.-R. Zheng, “A System-on-Chip and Paper-based Inkjet Printed Electrodes for a Hybrid Wearable Bio-Sensing System,” IEEE 34st Annu. Int. Conf. of the Engineering in Medicine and Biology Society(EMBC2012), Aug. 2012. Paper V: (invited) G. Yang, J. Mao, T. Hannu, and L- R. Zheng, “Design of a Self-organized Intelligent Electrode for Synchronous Measurement of Multiple Bio-signals in a Wearable Healthcare Monitoring System” IEEE Proc. Of the 3rd International Symposium on Applied Sciences in Biomedical and Communication Technologies (ISABEL 2010), pp. 1-5, Oct. 2010. Paper VI: G. Yang, J. Chen, F. Jonsson, T. Hannu, and L- R. Zheng, “A 1V 78uW Reconfigurable ASIC Embedded in an Intelligent Electrode for Continuous ECG Applications” 31st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2009), pp.2316-2319, Sep. 2009. (One of 15 best student paper finalists) Paper VII: G. Yang, J. Chen, Y. Cao, T. Hannu, and L- R. Zheng, “A Novel Wearable ECG Monitoring System Based on Active-Cable and Intelligent Electrodes,” IEEE Proc. of 10th International Conference on e-Health Networking, Applications and Services (HealthCom 2008), pp.156-159, Jul. 2008. List of Papers not included in the thesis, but related: Journal Papers: 1. L. Xie, G. Yang, M. Mäntysalo, L. L. Xu, F. Jonsson, and L.-R. Zheng ” Heterogeneous Integration of Bio-Sensing System-on-Chip and Printed Electronics” IEEE Journal on Emerging and Selected Topics in Circuits and Systems (JETCAS), vol.2, no.4, pp.672-682, Dec. 2012. 2. J. Chen, L. Rong, F. Jonsson, G. Yang, and L.-R. Zheng, “The Design of All-Digital Polar Transmitter based on ADPLL and Phase Synchronized ∆Σ Modulator,” IEEE Journal of Solid-State Circuits (JSSC) vol.47, no.5, pp.1154-1164, May 2012. Conference Papers: 3. (Invited) G. Yang, J. Chen, F. Jonsson, T. Hannu, and L- R. Zheng, “Bio-chip ASIC and printed flexible cable on paper substrate for wearable healthcare applications” ACM Proc. Of the 4th International Symposium on Applied Sciences in Biomedical and Communication Technologies (ISABEL 2011) Article No. 76, Oct. 2011. 4. Q. Wan, G. Yang, Q. Chen, and L-R. Zheng, “Electrical Performance of Inkjet Printed Flexible Cable for ECG Monitoring,” IEEE 20th Electrical Performance of Electronic Packaging and Systems, (EPEPS 2011), pp.231-234, Oct. 2011. 5. G. Yang, J. Chen, T. Hannu, and L- R. Zheng, “An ASIC Solution for Intelligent ix Electrodes and Active-cable Used in a Wearable ECG Monitoring System,” International Conference on Biomedical Electronics and Devices (BIODEVICES 2009), pp.209-213, Jan. 2009. 6. G. Yang, J. Chen, T. Hannu, and L- R. Zheng, “Intelligent Electrode Design for Long-Term ECG Monitoring at Home” IEEE Proc. of the 3rd International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth 2009), pp.1-4, Apr. 2009. 7. G. Yang, Y. Cao, J. Chen, T. Hannu, and L- R. Zheng, “An Active-cable Connected ECG Monitoring System for Ubiquitous Healthcare,” IEEE Proc. of 3rd International Conference on Convergence and hybrid Information Technology (ICCIT 2008), pp.392-397, Nov. 2008. Summary of the Appended papers: Paper I: G. Yang, L. Xie, M. Mäntysalo, J. Chen, T. Hannu, and L.-R. Zheng “Bio-Patch Design and Implementation Based on a Low-Power System-on-Chip and Paper-based Inkjet Printing Technology” IEEE Transactions on Information Technology in Biomedicine (T-ITB), vol.16, no.6, pp.1043-1050, Nov. 2012. A Bio-Patch prototype is demonstrated in this paper. Both the electrodes and interconnections are implemented by printing conductive nano-particle inks on a flexible photo paper substrate using inkjet printing technology. A Bio-Patch prototype is developed by integrating the SoC sensor, a soft battery, printed electrodes and interconnections on a photo paper substrate. The Bio-Patch can work alone or operate along with other patches to establish a wired network for synchronous multiple-channel bio-signals recording. The measurement results show that ECG and EMG are successfully detected in in-vivo tests. Contribution of the author: Original idea, most of the experimental work, and the draft of the manuscript. Paper II: G. Yang, J. Chen, L. Xie, J. Mao, T. Hannu, and L.-R. Zheng “A Hybrid Low Power Bio-Patch for Body Surface Potential Measurement” IEEE Journal of Biomedical and Health Informatics (J-BHI), 2013 (accepted). Three key technologies, including mixed-signal SoC technology, inkjet printing technology and anisotropic conductive adhesive (ACA) bonding technology, are employed to implement a wearable Bio-Patch prototype for body surface potential measurement. The electrodes, interconnections and interposer are implemented by inkjet-printing the silver ink x precisely on a flexible polyimide (PI) substrate. The reliability of printed silver traces on PI substrate is evaluated by static bending tests. ACA is used to attach the SoC to the printed interposer and form the flexible hybrid system. Measurement results show that the implemented Bio-Patch can concurrently detect ECG signals from 8 chest points. Contribution of the author: Original idea, most of the experimental work, and the draft of the manuscript. Paper III: G. Yang, J. Chen, F. Jonsson, T. Hannu, and L- R. Zheng, “A Programmable Bio-Electric SoC Sensor with 2-Wire Data Transmission for Scalable 14-Channel Synchronous BioSignal Measurement” Design, Automation & Test in Europe (DATE 2012) Mar. 2012. This paper presents the second tape-out of the Bio-Chip: a fully integrated SoC sensor for multiple bio-electric signals recording. It consists of an analog front end circuit with tunable bandwidth and programmable gain, a 6-input 8-bit SAR ADC, and a reconfigurable digital core. The SoC is fabricated in a 0.18-µm 1P6M CMOS technology, occupies an area of 1.5 × 3.0 mm2, and totally consumes a current of 16.7 µA from a 1.2 V supply. Circuit performances are reported, including measured gain and bandwidth, CMRR, and input referred noise. In-vivo tests are also performed to verify the SoC’s performance. EMG, ECG, and EEG signals are successfully recorded using the implemented SoC sensor. Contribution of the author: SoC design and implementation. All the test work for SoC electrical characteristics measurement and functional verification. All the in-vivo tests to measure bio-signal from volunteers’ body. The draft of the manuscript. Paper IV: L. Xie, G. Yang, M. Mäntysalo, F. Jonsson, and L.-R. Zheng, “A System-on-Chip and Paper-based Inkjet Printed Electrodes for a Hybrid Wearable Bio-Sensing System,” IEEE 34st Annu. Int. Conf. of the Engineering in Medicine and Biology Society(EMBC2012), Aug. 2012. This paper describes a hybrid wearable bio-sensing system, which combines the implemented SoC, and cost-effective Printed Electronics on a flexible paper substrate using inkjet printing technology. The electrodes are inkjet printed on photo paper substrate and connected to the SoC for measurements. In-vivo test results show that the quality of ECG signal sensed by printed electrodes is comparable with commercial electrodes, with noise level slightly increased The measurement results also indicate the SoC sensor operates smoothly with the printed electronics. The printed electrodes are flexible, light and thin, which make the final system comfortable for end-users. Contribution of the author: Original idea, parts of the experimental work, and the analysis of the measurement results. Paper V: G. Yang, J. Mao, T. Hannu, and L- R. Zheng, “Design of a Self-organized Intelligent Electrode for Synchronous Measurement of Multiple Bio-signals in a Wearable xi Healthcare Monitoring System” IEEE Proc. Of the 3rd International Symposium on Applied Sciences in Biomedical and Communication Technologies (ISABEL 2010), pp. 1-5, Oct. 2010. In addition to the digital core, a gain-bandwidth selectable analog front-end circuit should be designed and integrated on chip. It allows the circuit to accommodate different bio-potential signals. A SAR ADC is also included to convert the amplified bio-signal to a digital format. A full integration solution is achieved by integrating the three key parts together. In this paper, we present the architecture of the analog front end circuit, and the structure of the implemented SAR ADC. Post-layout simulation results show that the analog front end circuit is featured with a tunable gain, and for each selected gain, the bandwidth is also adjustable via external pins. Since during silicon chip manufacture process, device mismatch may lead to CMRR degradation, Monte Carlo analysis (Mismatch only) is performed to investigate the variation of the CMRR, where the simulation result is much closer to the real circuit performance. Contribution of the author: The digital core is the re-used IP core from the previous ASIC. All the design, simulation and lay-out work related to the analog front-end. Simulation and parts of the lay-out work for SAR ADC. The draft of the manuscript. Paper VI: G. Yang, J. Chen, F. Jonsson, T. Hannu, and L- R. Zheng, “A 1V 78uW Reconfigurable ASIC Embedded in an Intelligent Electrode for Continuous ECG Applications” 31st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2009), pp.2316-2319, Sep. 2009. After the function verification on FPGA, the proposed digital core is fabricated in a standard 0.18 µm CMOS technology (the analog front end circuit and A/D convertor are not included on chip). The ASIC occupies an area of 1.5 mm × 1.5 mm and is encapsulated in a CQFP-64 package. Experiments are made to measure the ASIC power consumptions under different supply voltages and operating frequencies. The minimum power consumption of 78 µW is observed with 1.0 V core voltage running with 6 MHz operating frequency. The tiny silicon size makes the ASIC possible and suitable to be embedded into an Intelligent Electrode, and the low power consumption makes it feasible for long-term continuous ECG monitoring. Contribution of the author: The digital ASIC implementation following the ASIC design flow (including VHDL code synthesis, floor-plan, place and route, DRC and post-layout simulation), PCB design for test board, function verification for the fabricated chip, power consumption measurement, and the draft of the manuscript. Paper VII: G. Yang, J. Chen, Y. Cao, T. Hannu, and L- R. Zheng, “A Novel Wearable ECG Monitoring System Based on Active-Cable and Intelligent Electrodes,” IEEE Proc. of 10th International Conference on e-Health Networking, Applications and Services (HealthCom 2008), pp.156-159, Jul. 2008. xii In this paper, for the first time, we propose the idea of Intelligent Electrode and Active Cable for wearable ECG monitoring systems. The block diagram of the digital core and the architecture of the Active Cable are illustrated. The structure of the command frame and the algorithm of the operation process are also presented. The digital core is embedded in the Intelligent Electrode and works as the controlling unit. Its function verification is performed using FPGA boards. Command packets and data loads are successfully transmitted between different nodes. Contribution of the author: The original idea, test method, all of the VHDL coding, all the experimental work, and the draft of the manuscript. xiii List of Acronyms ADC Analog to Digital Convertor AFE Analog Front End ASIC Application-Specific Integrated Circuit Ag/AgCl Silver/Silver Chloride BW Bandwidth CMOS Complementary Metal Oxide Semiconductor CMFB Common Mode Feed Back CMRR Common Mode Rejection Ratio CBIA Current Balanced Instrument Amplifier ECG Electrocardiogram, heart signal (also known as EKG) EEG Electroencephalogram, brain signal EOG Electrooculogram, eye signal EMG Electromyogram, muscle signal FPGA Field-Programmable Gate Array FFT Fast Fourier Transform FSM Finite State Machine GND Ground HCI Human Computer Interface IC Integrated Circuits ICT Information and Communication Technology IRN Input-Referred Noise SoC System on Chip RAM Random Access Memory RFID Radio frequency identification DPRAM Dual Port Random Access Memory SPRAM Single Port Random Access Memory SAR-ADC Successive Approximation Register Analog-Digital Converter SCL Serial Clock SDA Serial Data PWR Power REF Reference RF Radio Frequency PCB Printed Circuit Board PI Polyimide PHA Personal Health Assistant xiv Acknowledgements I want to express my most sincere thanks to my advisors Prof. Hannu Tenhunen and Prof. Li-Rong Zheng. They are always supportive and encouraging. I appreciate them for offering me this position, also for giving me this opportunity to conduct an exciting and independent research work covering Electrical Engineering and Biomedical Engineering. I appreciate Prof. Li-Rong’s ability to provide me an uplifting perspective which has inspired me throughout my research work. Many thanks to Prof. Hannu for his supports allowing me to join EIT ICT Doctoral Training Program. IC design is a complicated process. Project leaders and experienced colleagues show their supports in various ways. Special thanks to Dr. Fredrik Jonsson and Dr. Jian Chen, who are excellent circuit designers providing me with practical suggestion and support during circuit design and test. I am grateful to former colleague Ying Cao who helped me a lot on the DRC of my first ASIC. The constructive discussions with these intelligent colleagues have inspired me throughout my research, and I believe they are quite valuable in my future research work. Also many thanks to the powerful server Xcellent, who is a reliable friend with powerful computational capability, operating much faster than previous servers. I am grateful to colleague Li Xie, Matti Mäntysalo (Tampere University of Technology, Finland), Qiansu Wan, Jia Mao and Linlin Xu who contributed a lot on the bio-sensor integration, circuit layout, and sensor node miniaturization works by applying the printed electronics on the flexible substrates. Many thanks go to Ning Ma, Yang Li (Chang Chun University of Science and Technology, China), Awet Yemane Weldezion, Shahab Mokarizadeh, as well as the colleagues in the Department of Electronic Systems, who helped on the bio-signal extraction experiments, either making contribution as test volunteers or helping on the data recordings. I also want to thank all colleagues at KTH who make this Ph.D journey more enjoyable: Dr. Zhuo Zou, Dr. Yasar Amin, Dr. Liang Rong, Dr. Ana Lopéz Cabezas, Yi Feng, Jue Shen, Zhi Zhang, Zhiying Liu, Qin Zhou, Chuanying Zhai, Xueqian Zhao, Ming Liu, Shaoteng Liu, and Pei Liu. Special thanks go to Dr. Qiang Chen and Dr. Zhonghai Lu for their practical advice on the daily life in Sweden. Besides, I’d like to thank our secretary Agneta Herling and Alina Mutntenu for their administrative assistance in travelling and other issues. Thanks for the maintenance works provided by the people working in IT-Service Group. Special thanks to Eskil and Fredrik Liljeblad who rescued the IC design files from the crashed hard-drive. I could not imagine how much time and work would be spent on the redesign of the entire IC if the four-year design files were permanently lost. To my good friends, Dr. Huimin She and Dr. Botao Shao, together with whom I came to Sweden, the land of Nobel Prize, and at the same time started the journey of Ph.D study, thanks for your help and encouragement all along the way. Last, but not the least, I give my deepest gratitude to my wife Zhangwei Yu, who is an outstanding researcher in fiber optics, and I believe that one day in the future she will be an excellent scientist. For each of my published paper, there is an invisible contribution from her, reviewing my manuscript and giving me valuable comments. Sincere thanks for her sacrifices that she has made to let me achieve this goal. Also my son, Joey Jiuning Yang, like his name, he brings me lots of joys into my life. And to my elder sister in China, your endless concerns encourage me all the way, I will never let you down. I appreciate my parents for their endless love and irreplaceable supports. I am also grateful for their financial support which allowed me to have the college education for bachelor and master degrees. The research presented in this dissertation is financially supported by Vinnova (The Swedish Governmental Agency for Innovation Systems) through the Vinn Excellence centers program. Also I am grateful for the financial support from the China Scholarship Council (CSC) for my first three years Ph.D study. Geng Yang September 2012 Stockholm xiii Chapter 1 Introduction 1.1 Motivation and Pervasive Healthcare Consistent with the global population trend, many countries are facing aging problems. Chronic diseases are becoming the major causes of the death. In EU countries, the heart disease is the most common cause of death [1]. According to US National Center for Health Statistics, major chronic diseases such as heart disease, cerebrovascular disease, and diabetes account for 35.6% of death in US in the year 2005 [2]. Statistics show that over one-fifth of the world’s elderly population (aged 65 and over) lives in China in 2006, and the percentage of China’s elderly populace will be tripled to 24% by the year 2050 [3]. The prevalence of chronic diseases inevitably increases the total expenditure on healthcare and poses a grim challenge to the current healthcare systems worldwide [3]. Traditional healthcare and well-being services are usually offered within hospitals or medical centers. Citizens with chronic diseases as well as the people in post-surgery state need continuous monitoring of their health condition, especially the vital signs, until their health status becomes stable. Patients, as well as their families, also need to collaborate with their doctor and medical professionals to get informed about their states. Until now, the monitoring of the health condition of such people is usually accomplished within medical centers or hospital environments. As a result, measurements of vital signs and the corresponding diagnosis are carried out in controlled environments. However, this solution is costly, inefficient and inconvenient for the people with the need of routine checks, since the patients need to frequently visit the hospital, sometimes on a daily basis, or even worse, need a long-stay. There are huge requirements to move the routine medical check and healthcare services from hospital to the home environment, thus release the hospital beds and other limited resources to the people with urgent needs. Pervasive and personalized healthcare is a promising solution to provide efficient 1 medical services which could significantly lower down the healthcare cost. Pervasive healthcare may be understood from two perspectives [4, 5]. Firstly, it is the development and application of ambient intelligence and assistive technologies for human-centered healthcare and wellness management. Secondly, it tries to increase both the coverage and quality of healthcare by making healthcare available to anyone, anytime, and anywhere by removing location, time and other restraints. It takes the advantages of related advanced technologies in nowadays, such as advanced information and communication technologies (ICT) and materials technologies, etc, to implement the design concepts and integrate the healthcare seamlessly into everyday life [6-9]. Several examples towards pervasive, ubiquitous and personalized healthcare can be found in the following applications: Pervasive computing is employed to deliver elder care in [10], which is a system with many elements of pervasive computing to create intelligent living spaces for the elderly. A customized sensor node and software are realized in [11] for individual pervasive health applications, where the underlying hardware and software can be tailored for specific applications. The results reported in [12-15] show RFID technology applied in different medical care domains, which create smart medical care or smart hospital. Sensor technology is regarded as a key driver in the development of home healthcare. References [16-21]show different body sensors, intelligent sensors, and textile sensors, which are utilized in personal or pervasive healthcare. Electrocardiogram (ECG), electromyogram (EMG), electroencephalogram (EEG) and Electrooculogram (EOG) signals are the most common bioelectrical signals measured in clinical practice or for research purposes. They are bioelectrical signals generated by the activity of the heart, muscles, brain and eyes respectively, showing the characteristics of a dedicated physical or mental activity. The recording and monitoring of these bio-signals is used in a large number of applications: such as heart disease diagnosis, muscle activation analysis, eye movement detection and epilepsy diagnosis. In addition to the bioelectrical signals, body temperature, blood pressure, and blood glucose, etc. are also commonly measured vital parameters useful for assessing on the physical condition of a human being. There are several applications that might require continuous recording of bio-signals for relatively long time. Two illustrative examples of these applications are Holter monitoring, as shown in Fig. 1-1(a), and long term recording of multi-channel EEG signals for epilepsy diagnosis, as shown in Fig 1-1. (b). A Holter monitor is able to record ECG data for around 24-hour. Currently this is 2 (a) (b) Figure 1-1. (a) ECG Holter monitoring system. (b) High spatial density EEG system (TU Braunschweig). regarded as one of the most efficient methods to detect irregular heart activities. However, a measurement lead is required to connect one electrode to the Holter device. As a result, for the Holter monitor, typically 6 to 10 wires are required. For an EEG monitoring system in the traditional clinical area, more connecting cables are needed. For example, 20-40 wires are needed for a typical epilepsy EEG monitoring system. Additional electrodes and connecting cables could be added to the standard set-up when higher spatial resolution is demanded for a particular application. In both of the afore mentioned applications, a large number of cumbersome wires are worn around the body or head all day long, which incur tangling problems, moreover, make the recording process uncomfortable and inconvenient for the patients. It becomes an obstacle for the wider use of continuous monitoring system in the daily life. If the connecting cables bring too much inconvenience, users resist the adoption of wearable devices. In addition, cable noise and interference bring considerable technical challenges to the design of signal acquisition and conditioning circuits, which may involve extra needs of shielding or guarding techniques. 1.2 Work Performed Based on the above considerations, we propose a wearable bio-sensing system which 3 GP R Ne S/3 tw G ork Figure 1-2. Intelligent Electrode and Active Cable based wearable healthcare monitoring system. can be used for continuous home healthcare, as shown in Fig. 1-2, with an application scenario of multi-channel ECG measurement. A small microchip is embedded in each electrode, which endows the electrode with local intelligence. In this sense, the electrode is called Intelligent Electrode. With the integrated microchip, the Intelligent Electrode is able to obtain high quality bio-signal by conducting the bio-signal amplification, filtering, quantization, buffering, and transmission tasks directly on-site each electrode. Different from traditional approaches where the measurements are performed in a stationary or central device (centralized process), in this work, the bio-signal acquisition and transmission tasks are distributed onto each electrode (distributed process). By sharing a serial communication link (Active Cable), the Intelligent Electrodes can communicate with each other in a serial manner. The communication between Intelligent Electrodes are command based. The signal flow is illustrated in Fig. 1-2. Intelligent Electrodes detect the bio-signals on different sites of human body. These bio-signals are converted to digital format and gathered in one Intelligent Electrode, where these data are wirelessly transmitted to a smart phone or personal health assistant (PHA) for storage or real-time processing/display. According to requirements, the data can also be sent to a remote server or hospital/clinic center for storage, where further diagnosis can be made by the remote physicians. The concept of Intelligent Electrode and Active Cable are firstly reported in the attached Paper VII [22]. The advantages of this Intelligent Electrode and Active Cable based wearable healthcare system include: One-chip multi-sensor. The customized microchip is designed to sense a variety of bio-signals enabling the Intelligent Electrode to detect multiple bio-signals. 4 Drastically reduce the number of connecting cables. All sensing electrodes are connected by sharing a single Active Cable. As a result, user comfort is improved. On-chip implemented serial communication protocol enables the establishment of a body area network over a group of Intelligent Electrodes. To realize such a system, principally two core research challenges are present: 1) low-power and high-performance integrated circuits, 2) new heterogeneous integration technologies enabling sensor nodes/devices miniaturization. Accordingly, the contributions and focuses of this thesis can be summarized as follows: Design and implementation of the application-specific microchip and its function/ performance verification after chip fabrication. Design, fabricate and test the customized electrode patterns and evaluate their compatibility with the fabricated microchip through in-vivo tests. Performance evaluation of the printed conductive metal traces, interconnections and other patterns on flexible substrate. Propose feasible solutions and implement prototypes to demonstrate the concept of new wearable bio-sensing devices which could be used for next generation personalized/pervasive healthcare. 1.3 Author’s Contributions The main goal of this thesis is to propose, implement and evaluate the customized microchip and the relative integration technologies to enable the realization of wearable bio-signal monitoring systems for pervasive or personalized healthcare. To comply with the key requirement for wearable systems, the microchip needs to fulfill the stringent function requirements of bio-signal sensing, processing and transmitting. Great emphases are put on the design and implementation of the microchip. The building blocks are integrated on chip step by step. Starting with a digital controlling unit, followed by the design and implementation of analog read-out circuit and A/D convertor, and finally a full integration 5 solution is made by integrating the 3 building blocks on the same chip. Until now, two tape-outs have been made and the relative function/performance verification has been conducted through in-vivo tests. The author’s contribution can be concluded as follows: The configurable architecture of the digital controller is reported in [22]. For the first time, we propose the dual operating modes of the custom designed digital controller: Main-Mode (responsible for network management) and Sensing-Mode (responsible for bio-signal measurement). The author is responsible for the behavioral definition and the VHDL coding for the digital core’s two operation modes, including the serial communication protocol definition, RAM read/write control, RAM switch control, the definition of broadcast commands and unicast commands, the definition of command frame structure and data frame structure, the scheme of transmitting or receiving (interpreting) the command/data packet, the definition of serial port, ADC interface and wireless module interface. The author is also responsible for the verification of the above behavioral level codes on FPGA platforms. The concept and measurement results are reported in [22]. After the functional verification on an FPGA, the author moves forward with the application specific integrated circuits (ASIC) implementation. The author is responsible to implement the VHDL codes to a real digital controller by following the ASIC design flow. The author is also responsible for the following measurements when the microchip is fabricated. This is the first ASIC tape-out of this thesis work. UMC 0.18 µm CMOS technology is used and the total silicon area is 1.5 mm × 1.5 mm. The digital ASIC was fabricated in late 2008 and measurements were made in 2009. Design details are presented in [23] and [24]. ASIC’s measurement results are reported in [25]. In addition to the digital controller, a full integration solution needs analog front end circuit and A/D convertor. In paper [26], we report the circuit architecture of the analog front end and the SAR ADC. Post-layout simulation results, including CMRR, Monte-Carlo simulation, gain-bandwidth frequency response, and transient simulation of an ECG signal segment are also included. The author is responsible for all related works reported in paper [26]. With the accomplishment of this paper, the circuit is ready for the second tape-out. The second tape-out is a mixed-signal System on Chip (SoC), including three parts: 1) analog front end (AFE); 2) SAR ADC; and 3) digital core. The tape-out was made in late 2010 using UMC 0.18 µm 1P6M CMOS technology with a total size of 1.5 mm × 3 mm. Consequently, the measurements were made in 2011. Digital signal transmission experiments using the printed silver traces on paper substrates are successfully conducted. 6 Results can be found in [27] and [28]. A full report of the fabricated SoC’s performance is presented in [29], including circuit schematic, theory analysis, measurement results for circuit’s electric performances as well as experimental results obtained from healthy volunteers. The author is responsible for all the related works reported in [29], including the schematic, simulation and layout of the chip, as well as the experiments after chip fabrication. The performance of the printed electrodes is investigated in [30]. Test results show that the printed electrodes using inkjet printing technology have a good compatibility with the fabricated SoC sensor. The author’s contribution to this paper is mainly on the design of SoC’s measurement PCB board and applying it in the in-vivo test. Finally, system integration is performed by integrating the SoC on flexible substrates using the inkjet printing technology. A paper-based Bio-Patch prototype is presented in [31]. The SoC sensor is attached on a flexible photo paper substrate, where inkjet printing technology is employed to implement conductive electrodes and interconnections. ECG signal is successfully obtained from in-vivo test. Due to the physical limitation of the photo paper substrate, the author moves forward with a more stable polyimide (PI) substrate. A new Bio-Patch prototype is proposed and implemented in [32]. An alternative packaging method using printed interposer is also studied. The reliability of the inkjet printed traces is evaluated via static bending tests. Finally, multi-channel ECG signals are successfully recorded using the implemented Bio-Patch prototype [32]. 1.4 Thesis Outline The remaining parts of the thesis are organized as follows: Chapter 2 reviews the principle of bio-potential signal measurements and gives an overview of current electrode technology as well as current the-state-of-the-art designs in biomedical circuits and systems. Chapter 3 presents the proposed fully integrated SoC for multi-parameter bio-signal sensing. The circuit implementation and its electrical performances are demonstrated. Different bio-signals, including ECG, EMG, EOG, and EEG are recorded through in-vivo experiments. Chapter 4 shows the system integration and the implementation of Bio-Patch prototypes using the inkjet printing technology. The performance of the printed electrodes is investigated in this chapter. Two cases of hybrid integration of the SoC with printing technology on the flexible substrate are also demonstrated: one is based on photo paper 7 substrate; the other on the PI substrate. Finally, Chapter 5 concludes the thesis, and discusses the research works that could be continued in the future. This thesis is a combination of multidisciplinary research works that crosses Biomedical Circuits and System Design, Printed Electronics and Life Sciences. The main emphases of this thesis are on the microelectronic design and its related system integration, especially for the booming research areas such as Smart Sensors and Wearable Electronics for Pervasive Healthcare applications. 1.5 Beyond the Scope of this Work Biocompatibility is not studied in this work. The conductive electrodes can be developed with different materials, for example, conductive metals, conductive polymer, and textile electrodes etc. Various manufacture processes can be applied to fabricate these electrodes, for example, chemical etching, inkjet printing, screen printing etc. However, in this study, we focus on function verification of the implemented circuits, and investigate the possibility and feasibility of integrating the silicon SoC with Printed Electronics to form a hybrid bio-sensing system. The characteristics and biocompatibility of different electrodes are not considered in this work. Wireless transmission module is not integrated on the current Bio-Patch prototypes. Two solutions are available for the wireless communication: 1) a fully integrated solution with the radio circuits integrated on the same chip. In this case, the knowledge of integrated radio frequency (RF) transmitters in [33] and [34] can be used. 2) using an off-chip wireless module, where low energy Bluetooth [35] is a good candidate for RF solution. 8 Chapter 2 Bio-electrical Signal Sensing and Current Approaches 2.1 Introduction This chapter briefly introduces the nature of bio-electrical signals and the principles of bio-electrical measurements. The bio-potential signals discussed below mainly include the ECG, EMG, EOG, and EEG. They are widely used in clinical areas or non-clinical applications, and corresponding to the activity of respective organs or tissues. The bio-potential signals are generated due to the electrical activity of the living cells in human body, and these bio-signals can be traced back to a cell level [36]. The electrical potential generated by a propagating action through a nerve cell is very small and to record it with non-invasively methods is specially challenging. The bio-potential signal detected on the surface of the heart, brain or the human skin reflects the summation of the synchronous activity of thousands or millions of cells/neurons that have similar spatial orientation [36]. Organs or tissues, such as the heart, brain, muscles and eyes, conduct their function through endogenous electrical stimulation [37]. For example, the contraction of the cardiac muscles on the heart during a cardiac cycle generates a bioelectrical activity that give rise to the signal called ECG. Neuronal activity in the human cortex produce bioelectrical activity that allows the recording of the EEG signal. The activity of muscles, such as contraction and relaxation, generates bioelectrical potentials that when record produce an EMG signal. Eye movements cause the dipole by resting potential of the eyes when changing direction, resulting in a bio-potential signal known as EOG. Measurements of these bioelectrical signals from the body can provide valuable information for medical diagnosis. For instance, by analyzing an ECG waveform, abnormal heart beats or arrhythmias can be diagnosed. 9 TABLE I FEATURES AND APPLICATIONS OF BIO-POTENTIAL [39] [40] BIO-SIGNAL ECG EEG AMPLITUDE BANDWIDTH (MV) (HZ) 1-5 0.5-200 0.001-0.05 SELECTED MEASUREMENT ERROR SOURCE APPLICATIONS Motion artifact, Diagnosis of 50/60 Hz power-line interface arrhythmia. RF noise, 50/60 Hz Sleep studies, 0.5-100 seizure detection EMG 1-10 20-1500 RF noise, 50/60 Hz Muscle function, prosthesis EOG 0.4-1 DC-30 Motion artifact, skin potential Eye position track, sleep state. EEG signals are commonly used to interpret and identify epileptic seizure events by the neurologists. Nowadays, EEG is further investigated for brain-computer interface. EMG signals are helpful in assessing muscle function level for the athletes, also they can be used to drive the prosthesis for the disabled people. EOG signals can be used to trace the eye movement, eye position, sleep state analysis and the diagnosis of balance disorders. 2.2 Principles of Bio-potential Measurements Bio-potential recordings of ECG, EOG, EMG, and EEG are indispensible for clinical or research applications. With the help of modern bioinstrumentation technology, they can be recorded in a noninvasive way [38]. Table I lists the features, noise source of these bio-potential signals and their representative applications [39] [40]. The common features of bio-potentials are: Small amplitudes (from 10 µV to several mV), Low frequency range (from DC level to several kHz). The common problems encountered for the acquisitions of such signals are: 10 Environmental interference (power line, electromagnetic, etc.), Figure. 2-1. Electrodes placement for the bio-potential signals measurements. (a) ECG recording using right arm (RA), left arm (LA), reference (Ref), and six chest (Cen) electrodes. (b) EMG recording with two electrodes places on biceps and the reference electrode placed on the forearm. (c) EEG recording with selected electrodes locations from the standard 10-20 EEG lead system with an ear as reference. (d) EOG recording with electrodes placed up and down and sides of the eyes along with a reference placed behind an ear. Biological interference (motion noise, noise from skin and electrodes, etc.). Figure 2-1. illustrates the electrodes placement setup for the recording of specific bio-potential measurements. ECG Measurement The ECG signal is often recorded by placing electrodes on the torso, as shown in Fig. 2-1a. The ECG signal has an amplitude in the range of 1~5 mV and frequency contents ranging from 0.5 Hz to 200 Hz. The ECG is originated by the cardiogenic bio-electrical activity of the heart when the cardiac muscle depolarizes or repolarizes during each heartbeat. Usually multiple electrodes are used and ECG recording is widely accepted as a good way to measure and diagnose abnormal rhythms of the heart [41], particularly 11 abnormal rhythms caused by the damaged tissue of the heart. EMG Measurement Surface EMG is a recording of electrical activity produced by muscles. It is generated by muscle cells’ contraction. The signal can be used to evaluate muscles’ activation level. Measured surface EMG potentials range between 1 mV and 10 mV, depending on the contraction level of the muscle tissues, and frequency ranges from 20 Hz to 1000 Hz. Figure 2-1b shows the preferred electrodes placement for the EMG detection, where two electrodes (one pair of electrodes) are placed on the biceps, and a reference electrode is placed on far electrically unrelated skin. EEG Measurement Surface EEG is a summation of synchronous activity of a great number of neurons that have the same spatial orientation. Since the electrical potential generated by a single neuron is too small to be picked, the EEG signal reflects the brain activity where thousands or millions of neurons fire together. The electrical potentials captured on the scalp surface are characterized by extremely small amplitudes, sometimes too weak to be distinguished from the background noise, in the range of 1 µV to 50 µV, and the frequency ranges from 0.5 Hz to 100 Hz. Due to this reason, EEG measurement poses a great challenge to the read-out circuit. Figure 2-1c demonstrates the placement of the selected scalp electrodes for the standard 10-20 leads EEG system, where several sensing electrodes are located on the user’s forehead, and one ear is taken as the reference. This multiple leads system allows the diagnosis of seizure spikes as well as the feature extraction in sleep study [42, 43]. EOG Measurement EOG is the resulting signal from the eyeballs movements [44]. Electrical potentials are generated when the eyeballs move in a horizontal or vertical direction. These electrical signals can be detected via electrodes placed on specific positions around the eye. Corresponding electrodes placement setup is shown in Fig. 2-1d. The main applications are for ophthalmological diagnosis and eye movements tracking. The EOG signal is small (0.4 mV to 1 mV) and has low frequency components (from near DC to 30 Hz) [45, 46]. Amplifiers with a suitable gain and a good band pass response between near DC and 30 Hz are desirable. The principles of bio-potential recording techniques, especially the detailed electrode placement on the human body for ECG, EMG, EEG, and EOG measurements are 12 introduced above. The in-vivo bio-potential extraction experiments presented in Chapter 3 will follow the electrode deployment scheme presented in Fig. 2-1. 2.3 Current Approaches for Bio-potential Sensing This section contains a brief overview about electrode technology as well as the latest state-of-the-art circuits and systems for bioelectrical sensing. 2.3.1 Electrode Technology All bio-electrical signal acquisitions are made with the aid of electrodes which may be customized for specific needs [47]. The electrodes can be classified into different types according to their invasiveness. In this thesis, we focus on the least invasive one: the surface electrodes. Different types of electrodes are applied to interface the human body with the recording instruments/devices. According to the structure, the non-invasive, surface-attached electrodes can also be divided as: passive-electrode (no electronic inside), and active-electrode (off-the-shelf components inside). In this thesis we propose the idea of Intelligent Electrode (ASIC/SoC integrated inside), and based on this concept we further develop several Bio-Patch prototypes, combining the benefits of the integrated circuits and printed electronics, which will be discussed in Chapter 4. Passive Electrode Electrodes form an electrical interface between the human body and the sensing electronic circuits. They are regarded as a class of sensors that transduce bio-potentials from the human body to the measuring electronic circuits where the signals are sensed and processed. The conventional adhesive passive silver/silver chloride (Ag/AgCl) electrodes are commonly found in clinical and research applications today. The standard Ag/AgCl electrodes have been well-characterized and studied over many decades [48-50]. With proper preparation and electrodes deployment, they can provide good signal quality at a relatively low cost. However, drawbacks and several limitations associated to the use of Ag/AgCl gel electrodes still do exist. For the monitoring of chronic disease, conventional Ag/AgCl wet/pre-gelled electrodes can possibly produce irritation on the skin due to the long time attachment on a specific location on the skin (for instance, burn units [51] or neonatal care [52, 53]). In addition, the signal quality and system performance degrade when the 13 electrolyte gel dries out [48]. To address this problem, dry electrodes [48, 54] and textile electrodes [55, 56] are proposed as alternatives. However, most often textile electrodes present low conductivity limiting its applicability. For the dry contact electrode, comparing with its counter part - wet electrode, it is sensitive to motion artifact due to the comparatively unstable mechanical and electrical contact with the skin. Active Electrode In recent years, many attempts have been made to improve the traditional electrode technology. Several research groups are developing active electrode. The bio-signal amplification/buffering takes place at the electrode site, therefore an output signal with much better quality can be achieved. The relative low spread use of active electrodes, in contrast with the world wide spread standard electrode, is driven by novel applications requiring continuous bio-potential recording in very specific scenarios. Examples can be found in Fig. 2-2. The latest developments on active electrodes are presented as follows: Thomas Degen et al. presents a two-wired buffer electrode in [57], as shown in Fig. 2-2a. Yu Mike Chi et al. developed a non-contact active electrode for ECG/EEG applications [54, 58-60], as shown in Fig. 2-2b. Several attempts have been made to integrate active electrode in chairs (Fig. 2-2c) [61], and beds (Fig. 2-2d) [62] to obtain the user’s cardiac signal. A T-shirt based wearable ECG monitoring system is proposed in [63]. Although active electrodes are featured with improved immunity against environmental interferences [57], they are not used as often as expected yet. Firstly, a survey of published works shows that active electrodes are almost exclusively designed using off-the-shelf components, such as operational amplifiers [57]. The use of many off-the-shelf components leads to a cumbersome sensor size. Moreover, as the components carrier, the PCB itself is stiff. As a result, the characteristics of active electrode make the development of wearable (a) (b) (c) (d) Figure 2-2. The latest active electrode technology: (a) two-wired buffer electrode [57]; (b) non-contact active electrode [58]; (c) active electrode in chairs [61]; (d) active electrode in bed [62]. 14 measurement systems very challenging. Secondly, in addition to the signal leads, power leads VCC and ground GND are required for each active electrode, increasing the number of connecting cables in comparison with using passive electrodes. The resulting size and difficulty to integrate with garments are major limiting factors when considering implementing more complex circuits into the active electrode using the off-the-shelf components. 2.3.2 Latest State-of-the-art Biomedical Circuits and Systems With the developments in complementary metal-oxide semiconductor (CMOS) technology, fully integrated design has shown great advantages in many advanced medical applications, in terms of low-power consumption, high-level integration and high performance [64-67]. Comparing with its counterpart: the off-the-shelf components based active electrode, the IC solution is capable of integrating more complex circuit blocks with a much lower power consumption and smaller physical size. In the past decade, effort has been focused on developing biomedical circuits to enable physiological monitoring in novel scenarios as well as personalized healthcare applications. Since such biomedical circuits usually require battery operation or even being wirelessly powered, great emphasis has been paid on ultra-low power design, thus enabling the sensing circuits to operate for days or even months by a small battery or a wireless energy scavenger system. In addition, since those sensing devices or sensors need to be used in direct contact with the skin for relatively long time, more and more attentions are paid on sensor miniaturization. The final goal with such miniaturization is that by making the sensing circuit as small as possible, the resulting sensor should be unobtrusive and easy to be attached on human body or integrated into a wearable system. Examples can be found in battery-powered wearable or implantable medical devices: the bio-amplifier presented in [60, 68-71] can sense ECG activity; the ASIC reported in [72] can detect heart-rate variation; the implantable system in [73] senses the EMG signal for intelligent prostheses control; acquisition of EEG is performed in [74-77]; neural recording and neural tissue stimulator are presented in [78-86] and [87-93]. 15 16 Chapter 3 SoC Design, Implementation and in-vivo Tests In this chapter, the design and implementation of a Bio-electric SoC are presented. The performance of the fabricated SoC is assessed via in-vivo test. A serial communication protocol is implemented on chip, which enables the system to automatically detect the sensors attached to the serial link, thus realizes a self-managed network. By configuring the SoC to the specific gain and bandwidth, 4 different bio-signals, including ECG, EMG, EOG and EEG are successfully recorded. 3.1 SoC Architecture and Implementation The proposed SoC is a mixed-signal chip, which can be divided to 3 parts: 1) an analog front end (AFE), which is used to amply and process the weak bio-signal; 2) a successive approximation register analog to digital converter (SAR ADC), which is employed to convert the analog signal to a digital format; and 3) a digital core, which is responsible for controlling logic. The overall architecture of the Bio-electric SoC is presented in Fig. 3-1. The proposed SoC is developed step by step. We firstly start with the digital core part. It was designed and fabricated in 2008 and is the first IC tape-out of this thesis work. After Figure 3-1. Overall architecture of the Bio-electric SoC. 17 a satisfactory functional verification of this digital ASIC, we moved forward to the design of AFE and ADC, and finally a full integration was made by combining the 3 building blocks on the same chip. The mixed-signal SoC is the second IC tape-out of this dissertation work and was made in 2010. 3.1.1 Digital Core As the controlling logic, the Digital Core is responsible for digital bio-signal acquisition, buffer and transmission. In addition, the Digital Core also takes charge of communication between sensor nodes and the network management. The architecture of the proposed Digital Core is illustrated in Fig. 3-2. It includes the following features: Various interfaces: ADC interface (responsible for ADC control and digital data collection when the ADC output is ready.); RAM interface (use the on chip RAM to buffer the digital bio-signal); wireless interface (the measured bio-signals are gathered on one ASIC, and finally sent out via a wireless module). On chip RAM, which enables the Digital Core to buffer the digitized bio-signal. On chip implemented 2-wire serial transmission protocol, including the definition of the frame structure for broadcast commands and unicast commands. With this protocol, a body area network can be established and managed. Since a typical body area network requires at least two types of sensor nodes: one is responsible for sensing one or several specific bio-signals from human body, and the other Figure 3-2. Architecture of Digital Core. 18 type operates as a controller to manage the whole network. Our innovative contribution in this work is to propose and implement the two operating modes on the same chip. The advantages of this dual-operation mode architecture include: avoiding separate IC tape-out (also avoiding the use of two different types of chips in real applications), thus lower down the total cost and simplify the assembly process; easy configuration: the chip’s operation mode can be easily configured by setting the control pin to logic ‘1’ or ‘0’ using an external switch. The chip can be configured either to a Sensing Node (SN) or the Main Node (MN). The difference between the two operating modes exists in: SN mode: The main function of a SN is to sample, process and buffer the bio-signal. Each SN has a 4-bit address. The bio-signal from the subject’s skin is amplified and digitized. Subsequently, the digital bio-signal is picked up and buffered temporarily in SN’s on-chip RAM. If one RAM is full, the coming data are automatically switched to the second RAM. Before the on chip memory is full, the stored data will be periodically collected by the MN. MN mode: The main role of the MN is to manage the whole network using a variety of commands. SN-Chain Scan (SNCS) process is initiated when the system is powered up or reset. During SNCS process, all SNs in the chain are scanned, the address of each active SN is stored in the MN’s address buffer. Soon after SNCS is completed, bio-signal data collection process is initiated: the MN visits each SN in a serial manner. Collected data are buffered in a MN’s on-chip RAM. Before the storage is full, the MN forwards these data to the outside world via an external wireless transmission module, then it continues with the next round data collection. 19 (a) (b) Figure 3-3 (a) Microphotograph of the ASIC fabricated in 0.18 µm CMOS. (b) Measurement board, ASIC with packaged and the silicon die of the fabricated ASIC. The proposed Digital Core is implemented into a silicon chip by following the digital ASIC design flow: In brief, firstly, according to function specification, the Digital Core is divided into several function modules; each module is described in behavioral VHDL codes; the codes are simulated and verified in Modelsim; then followed by the functional verification using FGPA board; after that, the codes are synthesized using Synopsys and finally floor-plan, place-and-route are performed using SoCencounter, and generated GDSII file is sent to the foundry for ASIC manufacture. The Digital Core is fabricated in a 1P6M 0.18 µm mixed-mode CMOS technology occupying a total area of 1.5 mm × 1.5 mm. The microphotograph of the fabricated ASIC is shown in Fig. 3-3a. A CQFP-64 package is adopted to encapsulate the ASIC. A 6 cm × 5 cm circuit board is designed to verify the ASIC function blocks and measure the power consumption, as shown in Fig. 3-3b. The SNCS flow is described in Fig. 3-4a. The MN broadcasts a command packet containing a target SN address to the Active Cable. If the target SN is available, an acknowledgement frame is sent back immediately. Otherwise, if the target SN does not exist, no acknowledgement will be generated before MN’s timer overflow. Subsequently, the MN asserts the absence of the target SN, and continues with the next scan until all SNs are scanned. The current system supports 14 SNs and 1 MN. Fig. 3-4b illustrates the structure of the command packet. Fig. 3-4c shows the screenshot of SCL and SDA on an oscilloscope during SNCS. The ‘start’ and ‘stop’ commands for the implemented serial communication 20 Pwr Syn-Sample Bio-Data SN- Scan ... SN- Scan Collection Up CMD Command frame (31bit) SN Mem start Mem stop Broadcast Ack addr(4bit) addr(10bit) addr(10bit) Cmd(4bit) (3bit) (a) (b) (c) Figure 3-4 (a) Command-based SN-Chain Scan (SNCS) process. (b) Packet structure for the command frame. (c) Oscilloscope screenshot of the SCL and SDA. protocol are compatible with Philips I2C, while the definitions for the rest of commands, for instance, broadcast and unicast commands, are custom-designed. The average current consumptions of the digital core for the two operation modes are almost the same: 12.7 µA for the MN, and 12.6 µA for a SN from 1.2 V supply with a 1 MHz system clock. More measurement results can be found in the attached Paper VI [25]. 3.1.2 Bio-Electric SoC After the implementation of the Digital Core, we move forward to the AFE design. Since bio-potential signals exhibit weak amplitudes and low frequencies [94], these features bring considerable challenges to the bio-signal read-out circuit. A desirable front end circuit should have programmable gain and bandwidth, thus it can be configured to accommodate different bio-signals. The circuit is also required to fulfill the stringent needs for the bio-signal detection: high input impedance, low power and low noise. The circuit should be 21 (a) (b) Figure 3-5. (a) Simplified architecture of the 6-input 8-bit SAR ADC. (b) Timing sequence of one A/D conversion. able to pick up these weak signals from a noisy environment: amplify the target signal but effectively reject the unwanted noise. A programmable three-stage AFE is implemented in this design. The block diagram of the AFE is illustrated in Fig. 3-1. Stage1 is mainly composed of a current balanced instrumentation amplifier and a common mode feed back circuit. Stage1 provides a fixed gain of 7. The differential outputs of Stage1 are AC-coupled to Stage2. The close-loop gain of Stage2 is set to 12. Stage3 provides a tunable gain by tuning the equivalent capacitance of its inner capacitor bank via three external switches. The in-band gain of the Stage3 can be configured to eight different values ranging from 2 to 19. According to simulation results, the three-stage front-end circuit can totally provide a tunable close-loop gain ranging from 45 dB to 64 dB. The low-pass frequency of the front-end can be tuned to 80 Hz, 260 Hz, 400 Hz or 1 kHz by varying the on-chip load capacitance through the external switches. The three-stage gain-bandwidth tunable architecture allows the adjustment of the gain of the amplification chain to an optimum value, while avoiding saturation distortion. The AFE 22 Figure 3-6. Chip microphotograph with its building blocks marked. totally draws a current of 2.3 µA from a 1.2 V supply. Detailed schematics for each stage and the corresponding theory analysis are reported in Paper III [29]. A fully integrated solution needs an ADC [95, 96], since the digital data are required for data transmission and the backend digital signal processing. A SAR ADC is chosen as the quantization module due to its relatively good tradeoffs between power efficiency, conversion speed and resolution [97-99]. Fig. 3-5a illustrates the architecture of the implemented 6-input 8-bit charge-redistribution SAR ADC [100]. It converts an input voltage with binary-searching method. The circuit’s physical layout is implemented in a fully differential architecture. The Bandgap module of the ADC generates a temperature independent 800 mV reference (Vref) signal which severs as a reference voltage during A/D conversion. A MUX with 3 control pins is used to select one from six inputs to connect to the capacitive binary search array for A/D conversion. The input5 is reserved for the amplified bio-signal from the AFE; the other five inputs are available for future extension for other bio-signal sensing. A finite-state machine (FSM) controller controls the switches of the ADC capacitor array during each conversion step. The timing sequence is shown in Fig. 3-5b, where each A/D conversion is divided 8 steps, starting from the most significant bit and ending with the least significant bit. Each A/D conversion takes 30 clock cycles. The ADC sample rate is proportional to the external clock, for example, the ADC can provide a sample rate of 1 KS/s with a external clock frequency of 30 kHz. Similarly, a sample rate of 3.3 KS/s can be obtained with a external clock of 100 kHz. The ADC takes an area of 480 µm × 580 µm and consumes 1.8 µA from a 1.2 V supply (30 kHz clock frequency). 23 The Bio-electric SoC is accomplished by packing the Digital Core, AFE, and ADC on the same chip. The SoC is fabricated in a standard 0.18-µm Mixed-Mode 1P6M CMOS process and totally occupies 1.5 mm × 3 mm silicon area with a operating voltage ranging from 1.2 V to 3.3 V. The chip microphotograph is shown in Fig. 3-6. A printed circuit board (PCB) is designed for SoC electric performance evaluation and in-vivo tests. A photograph of the measurement PCB is shown in Fig. 3-7. The in-vivo experiments performed and shown in the following parts of this chapter are based on this PCB. In order to accommodate different bio-electric signals, the AFE is designed to have tunable gain and bandwidth. An intuitive comparison is made in Fig. 3-8. The AFE gain-bandwidth post layout simulation result is plotted in Fig. 3-8a, and gain-bandwidth measurement result from a fabricated SoC is plotted in Fig. 3-8b. It can be observed that the measured gain and frequency response are in line with the circuit simulation results. The only difference observed that the measured gain is slightly smaller than the simulation result, for example, the measured largest gain (63 dB) is a little bit smaller than the simulated value (64 dB). This phenomenon may be due to the attenuation of the unit buffer located at the output port of the AFE (the function of this unit buffer is to increase the AFE driving capability). As a result, the measured gain is slightly smaller than the simulation result. The gain or bandwidth can be independently adjusted via external pins. Totally 8 different gain, ranging from 45 dB to 63 dB, can be selected (three of them chosen for Figure 3-7. 24 SoC measurement PCB. 80 BW adjustment through BW switches 70 Gain = 64 Gain = 59 Gain (dB) 60 Gain = 51 50 40 30 20 10 0.1 1 10 100 Frequency (Hz) 1000 (a) 65 63dB; 0.37~850 Hz Gain (dB) 60 55 57dB; 0.3~1.3k Hz 50 45 40 35 30 0.1 49dB; 0.35~1.5k Hz BW Adjustment Gain:62 dB; BW:0.30~200 Hz Gain:62 dB; BW:0.25~160 Hz Gain:61 dB; BW:0.33~85 Hz 1 10 100 Frequency (Hz) 1000 (b) Figure 3-8. (a) AFE Gain-bandwidth simulation result. (b) AFE Gain-bandwidth measurement result. illustration in Fig. 3-8a and Fig. 3-8b). For each selected gain, four user-programmable 3-dB bandwidths can be obtained using bandwidth switches. The upper part of Fig. 3-8b shows the case of 3-dB bandwidth adjustment with an almost consistent gain with the low cut-off frequency around 0.3 Hz and the high cut-off frequency adjustable from 85 Hz to 850 Hz. Due to its weak amplitude, the bio-signal is prone to be contaminated by different noise sources. Practical problems may be encountered during bio-signal recording process which 25 arise from physiological and environmental electronic noise sources. Physiological sources of interference are motion noise, muscle noise, etc. Electrical interference arises from the usual sources: 50-Hz power lines, and radio frequencies (RF) noise which are electrically or magnetically induced interference [40]. An effective approach to mitigate this noise influence is to apply a differential architecture in the amplifier design. Fully differential architecture is employed for the AFE design, which enables the effective rejection of the common mode interference. As a result, any common mode signal fed to the AFE input ports will be effectively removed, while the differential signals at the two sites will be amplified. 50-Hz power-line interference is the most commonly observed noise in the measured bio-signals [71, 101]. Since the power lines noise shows itself as a common mode signal on the two input ports, it will be removed by the circuit subtraction. The circuit noise is analyzed by connecting the buffer output to a signal analyzer: HP Dynamic Signal Analyzer 35670A. In order to measure input-referred noise (IRN), AFE’s two input terminals are shorted to ground and the output noise is measured with the signal analyzer. The measured noise divided by the measured gain at the closet frequency point equals IRN. Fig. 3-9 shows the IRN spectrum density. The noise floor of -145 dB Vrms/√Hz (56 nV/√Hz) is observed. It can also be observed that the AFE’s low frequency noise is mainly dominated by the flicker noise. At higher frequencies, the noise level approaches the thermal noise limit of the circuit’s differential input pair. The SoC’s total power consumption is 20 µW from a 1.2 V battery. The power breakdown is shown in Fig. -100 -120 -140 -160 -180 0 10 1 10 10 2 Figure 3-9. Input-referred noise of the analog front-end. 26 Figure 3-10. SoC power breakdown. 3-10, where the power consumptions for Digital Core’s two operating modes (SN mode and MN mode) are almost the same, around 15 µW from 1.2 V supply. 3.2 SoC in-vivo Tests In order to evaluate the SoC’s performance, the fabricated SoC is verified in the in-vivo tests by recording real bio-electrical signals from a healthy volunteer. Figure 3-11 illustrates the gain-bandwidth setup which are applied to the following in-vivo tests. The gain of 49 70 65 63dB;0.37Hz~850Hz 60 57dB;0.25Hz~190Hz Gain (dB) 55 57dB;17Hz~1.5KHz 50 45 49dB;0.35Hz~1.5KHz 40 EEG: 63 dB; 0.37~850Hz EOG: 57 dB; 0.25~190Hz EMG: 57 dB; 17Hz~1.5KHz ECG: 49 dB; 0.35Hz~1.5KHz 35 30 0.1 1 10 100 Frequency (Hz) 1000 Figure 3-11. Gain and bandwidth setup for in-vivo tests. 27 TABLE II SOC SETUP FOR DIFFERENT RECORDING APPLICATIONS Gain switch set BW switch set Gain of AFE BW of AFE ECG (0.5-150 Hz) ‘010’ ‘000’ (0.1 – 3 mV) 49 dB 0.35 Hz-1.5 kHz EMG(14-350 Hz) ‘100’ ‘000’ (0.01 – 2 mV) 57 dB 17 Hz-1.5 kHz EOG (0-30 Hz) ‘100’ ‘111’ (0.1 - 1 mV) 57 dB 0.25 Hz-190 Hz EEG (0.5-40 Hz) ‘111’ ‘011’ (10 µV - 0.1 mV) 62 dB 0.37 Hz-850 Hz Bio-signal dB with bandwidth between 0.35 Hz and 1.5 kHz is for ECG recording; the gain of 57 dB with bandwidth from 17 Hz to 1.5 kHz is for EMG measurement; the gain of 57 dB with bandwidth from 0.25 Hz to 190 Hz is for EOG signal; the gain of 63 dB with bandwidth from 0.37 Hz to 850 Hz is for EEG signal. Table II summarizes the amplitude and frequency characteristics of the target Figure 3-12. Photograph for ECG measurement. 28 bio-potential signals, and the SoC parameter setup for a specific recording application. Tyco Arbo H124sg pre-gelled adhesive electrodes (24 mm diameter) are employed to interface the SoC with the human body. The electrode placement for each individual test follows the electrode deployment instruction illustrated previously in Fig. 2-1. ECG measurement The ECG measurement is performed with two electrodes placed on a subject’s torso, and a third electrode as a ground electrode placed in the middle of subject’s chest. The measurement PCB is powered by battery with SoC output connected to an oscilloscope. An ECG wave with a clear QRS complex is visible on the oscilloscope, as shown in Fig. 3-12, where the oscilloscope is stopped for photographing. It should be noted that the captured waveform is raw data, since the 50-Hz power line noise is effectively rejected by the fully differential AFE, no extra digital signal processing is needed in this work. EMG measurement For EMG measurement, the electrode pair is placed on the subject’s biceps of right arm. EMG signal is detected when the subject contracts and relaxes his biceps. Figure 3-13 presents the measured 4-second EMG signal and its Fast Fourier Transform (FFT) of a selected signal segment. The spectrum shows that most of the EMG energy locates in the frequency range from DC to 300 Hz; and the dominant energy is between 10 Hz to 150 Hz, which is in line with [102]. 12 EMG FFT 10 1.0 Amp (mV) Amp (V) 8 0.8 6 4 2 0.6 0 1 2 Time (s) 3 4 0 200 400 600 Frequency (Hz) Figure 3-13. Measured EMG signal and its FFT. 29 Figure 3-14. Photograph for EOG measurement. EOG measurement For the EOG signal detection, the sensing electrode is placed outside an outer canthus, which provides a straightforward detection of eye horizontal movements. Figure 3-14 shows a captured EOG signal, where the subject is asked to read a book line by line. Sequences of small saccades (denoted as ‘r’ in Fig. 3-14) occur when the eyes move to the right over the words. A large saccade (denoted as ‘L’ in Fig. 3-14) is observed when the eyes move back to the left to the beginning of the next line. EEG measurement EEG Test Recording Electrodes Ground Electrode Figure 3-15. Photograph for EEG measurement. 30 EEG signal (V) 1.0 eyes blinks 0.6 0.4 5 10 15 20 Time (s) 25 30 35 40 0.005 Alpha Wave (Eyes closed) Eyes open Eyes closed 0.70 0.004 0.65 0.60 35 0.75 36 37 No Alpha Wave (Eyes open) 0.70 FFT Amp (V) EEG signal (V) EEG signal (V) eyes closed 0.8 0 0.75 eyes open 15 0.003 38 0.002 0.001 0.65 0.60 Dominant Rhythm 10 Hz 16 Figure 3-15. Photograph for0.000 EEG measurement. Time (s) 17 18 10 20 30 40 50 60 Frequency (Hz) Figure. 3-16. Measured EEG signal using the implemented SoC from the frontal electrode with eyes blinks, and signal segments for eyes-open and eyes-closed cases in time-domain, frequency-domain. EEG signal is difficult to detect, since its amplitude is quite small, sometimes even smaller than the background noise. Moreover, EEG signal is difficult to interpret, since it represents the comprehensive electrical activity of thousands or millions of neurons transmitted via brain tissues to the scalp [43]. However, certain features can be interpreted. Analysis of the EEG signals in frequency spectrum can reveal the signal power in different frequencies, which are produced during various status of the brain, such as in sleep status or wake up status [42]. There is even notable difference shown on the EEG signals in frequency domain when the subject keep his/her eyes open or closed. A photograph of EEG measurement scenario is presented in Fig. 3-15 using the SoC measurement PCB. An eye blink experiment has been performed, where the subject is asked to blink his eyes for several times before keeping his eyes closed for around 20 seconds. Fig. 3-16 demonstrates the recorded EEG waveform along with selected segments for eyes-open and eyes-closed experiments in time-domain and frequency-domain. Alpha wave (in frequency range of 8~13 Hz) is usually used as an indicator to verify if EEG signal is successfully recorded using a specific recording device. The Alpha wave shows up when the subject’s body and mind are at rest with eyes closed [103]. As shown in Fig. 3-16, the large spikes in the time-domain EEG waveform (upper part 31 of Fig. 3-16) correspond to the subject’s eye blinks [103]. Two EEG signal sections are selected and enlarged (lower left part of Fig. 3-16). They correspond to the recording of three-second time slot for “Eyes closed” and “Eyes open” cases. From the time-domain waveform, it can be clearly observed that the Alpha wave appears when the subject’s eyes are closed and disappears when the eyes are open. The FFT results (lower right part of Fig. 3-16) of the “Eyes closed” and “Eyes open” segments further demonstrate and prove the experimental findings regarding the Alfa wave. When the subject’s eyes are closed, a noticeable increase in power of Alpha band with frequency around 10 Hz is observed. While for the eyes open case, the subject is stimulated by the environment, as a result, the Alpha wave is abolished. Since the experiments are taken in a typical electrical engineering laboratory, a slight sign of 50 Hz power-line interference is observed in the FFT result of the captured EEG signal. 3.3 Conclusion A fully integrated low-power Bio-electric SoC is presented. The SoC sensor is featured with programmable gain and bandwidth to accommodate a variety of bio-signals. Comparison between the simulation results and the measurement results is made. It can be observed that the measured gain-bandwidth performance is in line with the simulation result. The SoC is fabricated in a 0.18-µm standard CMOS technology. The implemented SoC occupies a silicon area of 1.5 mm × 3 mm and totally consumes 20 µW from a 1.2 V supply. Totally 8 different gain, ranging from 45 dB to 63 dB can be achieved. On chip implemented serial communication protocol enables the establishment of a body sensor network over a group SoC sensors. SoC’s performance is verified through in-vivo tests. ECG, EMG, EOG, and EEG signals are successfully recorded using the implemented SoC sensor. Brief analyses of the measured biosignals are also made which further prove the performance of the implemented SoC. The measurement results show that the proposed SoC can serve as a versatile bioelectric sensor and can be applied in many healthcare and wellness related applications. 32 Chapter 4 System Integration and Bio-Patch Implementation The in-vivo tests presented in the previous chapter using the implemented SoC are conducted by applying the following commercially available electrodes, Tyco Arbo H124sg pre-gelled adhesive electrodes, on the surface of skin, and connecting the electrodes to the sensing SoC using lead wires. Although the commercial electrodes provide a relatively stable mechanical and electric connection, the size for electrode size and snap button is fixed. Since the implemented SoC has tunable gain and bandwidth, it allows the interface with customized electrodes with a much smaller size. Patterned electrodes and customized interconnections are needed for sensor miniaturization. In this chapter, characteristics of printed electrodes are investigated. Electrode optimization is made to find a suitable electrode pair with proper size and distance compatible with the fabricated SoC. Moreover, since the sensor needs to be applied on human body, user comfort is an important issue. For printing the electrodes, a flexible substrate is employed as circuit carrier, which can be easily bent to fit the curve of human body. High resolution inkjet printing technology is utilized for printing the electrodes, the interconnections as well as other patterns directly onto flexible substrates. In this way two prototypes of Bio-Patch have been successfully implemented on photo paper substrate and polyimide substrate respectively. 4.1 Printed Electrode Inkjet printing technology is employed in this design to print conductive electrodes and circuit board. Compared with conventional PCB manufacturing, the inkjet printing is more cost efficient and environmental friendly. The advantages of utilizing this technology include the following: 33 1) Less processing steps. Inkjet printing tech supports roll to roll massive production. The silver electrodes and all the interconnections can be made at the same time on the same substrate. Less process steps means time and energy efficient. 2) Easy assembly. Users or healthcare providers do not need to snap the commercial electrodes on/off each patch, thus save maintenance/operation time. 3) From environment point of view, compared with conventional PCB manufacturing, inkjet printing tech is more environmental friendly, since it does not involve masks or etching steps, thus avoids the use of harsh chemicals. Also, as a print-on-demand and digital printing method, material waste is minimized considerably. 4) Electronic devices miniaturization and steps towards new packaging technology. Inkjet-printing of nano-metal particles can be used to connect chips to package (e.g. System-in-Package) or directly print circuits on boards (e.g. Chip-on-Board). More compact patch/sensor can be achieved due to the thin profile of printed electrodes/interconnections. The smaller/more unobtrusive the patch/sensor is the more convenient and comfortable for the user to use. 5) A step towards full integration of hybrid biomedical systems. By utilizing the digital inkjet printing tech, the hybrid system can take advantages of both the fully integrated SoC and the emerging Printed Electronics. In this research work, the digital inkjet printing technology is used to fabricate the printed electrodes. It is a contact-free manufacturing technology, and does not involve any complicated process steps, so passive elements can be easily fabricated in a short time at relatively low cost [104-106]. Nano-particle sliver ink is directly printed on the flexible substrate to form electrodes and circuit interconnections [107]. The printed structure is sintered by heat in an oven, so that the nano-particles join together and form a continuous structure that enables formation of metal clusters/plate. In addition, sintering step can effectively remove the solvents and other additives present in the ink by thermal decomposition and evaporation [108]. HP Advanced Photo Paper is selected as the substrate, since it is low cost, environmental friendly and disposable after use. Moreover due to its physical features, such as it is flexible, with smooth surface, good water resistivity, it attaches well to the human body with respect to the skin convexity. 34 Pair A Printed Electrode Pair Pair B Pair C Pair D Pair E Figure 4-1. Printed electrodes using silver nano-particle ink on a photo paper substrate. Paper based printed electrode can be considered as a potential alternative to the commercial electrode, since it is featured with low manufacture cost and disposability after use. In order to investigate the compatibility of the printed electrodes with the implemented Bio-electric SoC, several pairs of electrodes are printed on the photo paper substrate. Figure 4-1 shows the fabricated samples of inkjet-printed silver electrodes and connection traces. The diameter of the electrode is 15 mm, and the electrode distance varies from 6 cm to 14 cm. The performance of the printed electrode pairs is investigated by connecting them to the SoC’s measurement PCB and examine the output signal quality on the oscilloscope. The electrode pair with electrode distance of 12 cm is selected for demonstration, as shown in Fig. 4-2. The in-vivo tests presented below are based on this electrode pair. Figure 4-3 shows the cross section view of the printed electrode. The thickness of the paper substrate is approximately 250 µm. The thickness of silver layer is approximately in the range of 2 ~ 5 µm. It is hard to clearly distinguish the interface between silver and paper layers, due to the absorption of the ink to the paper and evaporation during sintering. 15 mm Printed Sliver Trace Printed Electrode 12 cm Figure 4-2. Printed electrode pair with electrode distance of 12 cm on a photo paper substrate. 35 Silver trace 2~5 um Photo Paper 250 um Figure 4-3. Cross section view of the printed electrode. The performance of the printed electrodes is firstly evaluated by connecting them to the measurement PCB through the differential inputs of the SoC’s instrumentation amplifier. Figure 4-4 shows the photograph of ECG in-vivo test scenario using the printed electrodes and the standard measurement PCB. In this figure, the oscilloscope is stopped with captured ECG signal on the screen; the printed electrodes are intentionally held outward of the chest (instead of towards the chest) for demonstration purpose. During monitoring, the electrodes are placed towards the chest, and an elastic bandage is used to apply mild pressure on the electrodes, so they are well attached to the skin. A clean ECG waveform is obtained from the output of SoC’s analog front-end and it can be easily observed on the oscilloscope Figure 4-4. ECG test with paper based printed electrodes. 36 Figure 4-5. EMG test with paper based printed electrodes. shown in Fig. 4-4. For the EMG measurement, another electrode pair is placed on the subject’s forearm. The bandage is applied to fasten the electrode to the arm to minimize the motion artifact. EMG signal is detected when the subject contract and relax her hand, as shown in Fig. 4-5. More investigation and measurement results regarding how the electrode distance affects the amplitude of the detected signal are reported in attached Paper IV [30]. It can be observed that all the characteristics of the measured ECG and EMG waves are clearly visible, without any digital signal processing. The measurement results prove that the printed electrodes are adequate to detect the weak bio-potential signals and can operate smoothly with the SoC sensor. 4.2 Paper based Bio-Patch After the performance of the printed electrodes has been evaluated, we move forward to a full integration solution by integrating the SoC sensor and the printed electrodes on the same substrate. 37 Figure 4-6. Inkjet printed circuit board and patterned electrodes on a photo paper substrate. Silver ink is directly printed on a photo paper substrate to make interconnections, thus a paper-based circuit board is formed. Figure 4-6 shows the printed circuit board and patterned electrodes on a photo paper substrate. A prototype of the proposed Bio-Patch is implemented on a photo paper substrate, as shown in Fig. 4-7, with a size of 6 cm × 8 cm. It is composed of three parts: a pair of printed electrodes, a soft battery, and the bio-electric SoC. A flat Enfucell battery is adopted in the Bio-Patch implementation. The soft battery is flexible and suitable for low power applications, such as micro-sensors, printing and disposable applications. It provides a voltage source of 3.0 V with a battery capacity of 10 mAh. By attaching it on a subject’s chest, the proposed Bio-Patch can work as one channel ECG sensor. The collected digital ECG signal can be transmitted to the outside world via a Figure 4-7. Bio-patch implemented on a photo paper substrate with printed electrodes, interconnections, bio-electric SoC and a soft battery. 38 Figure 4-8. Crack generated in the silver trace on photo paper substrate by large folding. wireless module. Bluetooth or custom designed wireless transmitter can be employed for wireless transmission. A novel all-digital polar transmitter is investigated which is implemented and tested on another chip. The measurement results are reported in [33]. The Bio-Patch can work alone or operate along with other patches to establish a wired network for synchronous multiple-channel bio-signals recording. The measurement results show that ECG and EMG are successfully measured in in-vivo tests. More results are reported in attached Paper I [31]. 4.3 Static Bending Tests A common concern encountered by almost all the inkjet printed circuit built on flexible substrate is the crack probably generated when the substrate is under extreme bending. The crack leads to electric path discontinuity (open circuit). For the photo paper based Bio-Patch, during the in-vivo tests, the electrical performance is consistent even with a large bending (but not folded) to fit the body convexity. However, sharp edge-folding on paper substrates creates deformation and crack on the substrate. The metal trace becomes split apart as well. Figure 4-8 presents the photographs of cracks in the silver trace, generated when the substrate is folded. 39 (a) (b) Figure 4-9. Static bending tests. (a) Test samples printed on photo paper substrate. (b) Test samples printed on PI substrate. To assess the reliability of the printed sliver traces on the flexible substrate, static bending test is performed. 2 types of flexible substrate, photo paper substrate and polyimide (PI) substrate respectively, are selected for test. 3 identical samples are printed on photo paper and PI with a length of 4.5 cm and width of 2 mm, as shown in Fig. 4-9. During tests, a multimeter is used to measure the DC resistance of the printed silver trace. We gradually decrease the bending radius of the printed samples to examine if metal crack or peel-off could occur under any condition. The measurement results of the bending tests performed on photo paper substrate are summarized in Fig. 4-10, where the DC resistance varies as a function of bending radius. It Resistance (ohm) 10 9 8 Sample1 Sample2 Sample3 7 6 10 9 8 7 6 5 4 3 2 1 0 Bending Radius (cm) Figure 4-10. Resistance variation versus bending radius on photo paper substrate. 40 Figure 4-11. Static bending test for PI based silver trace, no crack occurs under extreme folding. can be observed that the resistance of the 3 printed samples remains stable (resistance variation less than 1%) when the bending radius is decreased from 10 cm to 1.3 cm. However, all 3 samples crack (the resulting resistance exceeds the maximum range of the multimeter) when the photo paper substrates are bended to 1 cm radius. For the PI based silver traces, throughout the entire experiment, no metal crack or peel-off occurs, and the samples keep an almost consistent resistance even when the radius is decreased to 4 mm (which means the sample is spiral-wound of 4 turns, a quite large bending considering with the sample size). Finally, we make an extreme bending test by folding the sample and pressing hard on it to form a sharp edge, as shown in Fig. 4-11. From the figure, it can be observed that the connection of the silver trace on PI substrate keeps intact with the resistance of around 30 Ω. The measurement results indicate that the printed silver traces on PI substrate exhibit remarkable advantages over the ones based on photo paper substrate, in terms of reliability and flexibility. It can be reliably employed in the fabrication of conductive electrodes, Active Cable and printed flexible circuit board. We move forward with the wearable healthcare monitoring system integration using PI substrate. In addition, a novel packaging method is applied on PI substrate to achieve more compact patch/sensor size, which will be presented in the following parts of this chapter. 41 4.4 Polyimide (PI) based Bio-Patch Photo paper based printing electronics shows its advantages in terms of low cost, environmental friendly and disposable after use. However, due to its physical nature, the paper substrate breaks when folded. As a result, cracks are generated in the printed metal trace producing an open circuit. Bending test results presented in the previous section show that PI based printed silver traces have better reliability. PI substrate also exhibits good flexibility and suitable for wearable devices. In addition, another drawback for the implemented photo paper based Bio-Patch is that the size is relatively large for an on-body sensor (6 cm × 8 cm shown in Fig. 4-7.). From IC package point of view, since currently many pins are for test and debug purpose, package JLCC 84 (3 cm × 3 cm) is chosen. In many advanced medical applications, such as wearable healthcare devices, a desirable bio-sensor should be unobtrusive, light with tiny size [109]. We realize the current SoC package size is the bottleneck for the further miniature of the Bio-Patch. This problem can be solved by employing a smaller IC package (such as QFN or SOIC package). Alternatively, we try to deal with it in another approach instead: using the inkjet printing technology presented in [110] to produce electrical interconnections on an interposer and directly attach the flip SoC chip on the interposer, thus avoid the use of a rigid ‘package’. With this method, it is possible to implement the patch with a much smaller size at a lower cost, thus make the sensor node suitable for on-body deployment. In the following parts of this section, we present a Bio-Patch prototype implemented on PI substrate. The PI based Bio-Patch prototype is formed by assembling the inkjet printed interposer lead-frame, conductive electrodes, and printed circuit board on the PI substrate. Anisotropic conductive adhesive (ACA) is applied in this prototype. It is fabricated by dispersing conductive particles with tiny diameter in the adhesive. It has been widely used in current flip chip packaging applications [111, 112]. ACA provides both electrical and mechanical connections between chip pads and the substrate. The distinct advantages of using ACAs include reduced package size and thickness, improved environmental compatibility, and lower assembly temperature [113]. In addition, the process costs are lower compared with traditional bonding/soldering approaches, due to fewer processing steps [114]. Furthermore, ACA can be used in fine pitch applications as small as 40-um pad width and 80-um pad pitch [115]. Static and cyclic bending tests performed in [115] also indicate that the ACA joints are featured with sufficient reliability for consumer electronics. 42 (a) (b) Figure 4-12. Assembly process with ACA. (a) Silicon chip alignment. (b) Thermo- compression bonding. In this design, ACA is employed to bond the Bio-electric SoC chip onto the PI interposer to form a sensor node. A cross-section view of the overall flip chip assembly process using ACA is shown in Fig. 4-12. In the initial stage, the adhesive is not conductive in any direction, as conductive particles do not touch each other due to their tiny particle size and low density dispensed in adhesive, as shown in Fig. 4-12a. When pressure and heat are applied, the conductive particles between the chip pad and substrate are deformed or breached, so the electrical contact can be made in vertical direction (perpendicular to the plane of the die), while in the horizontal direction it still keeps non-conducting, as shown in Fig. 4-12b. During the assembly process of this work, a thin layer of ACA is applied over the printed leads on the interposer. The flip chip is then aligned and attached onto the interposer with a Fine Placer, which also provides proper pressure and heat required for this packaging process. The inkjet printed interposer is shown in Fig. 4-13a, The interposer with the attached SoC using ACA is shown in Fig. 4-13b. In order to ensure the electrical and mechanical contacts between the die and substrate, a protecting mold (medical compatible epoxy from EPOXY-Tech) is applied over the flip chip area, as shown in Fig. 4-13c. The assembled interposer has a total dimension of 18 × 18 × 0.03 mm3. The physical size of the interposer (a) (b) (c) (d) Figure 4-13. SoC bonding and packaging. (a)Interposer on PI substrate. (b) Flip chip bonding with ACA. (c) Protecting mold. (d) Sensor node assembly. 43 Figure 4-14. Prototype of PI based Bio-Patch. is much smaller and thinner than a standard JLCC-86 package (30 × 30 × 3 mm3 , which was used as the chip container in previous prototype). The integration of the tiny-size SoC on the printed interposer makes the sensor node miniaturization possible. Finally, the interposer is flipped and attached to a larger PI mid-layer using ACA to form a sensor node, where several passive components are applied to configure the chip to a proper working condition, as shown in Fig. 4-13d. Figure 4-14 shows the final assembled Bio-Patch prototype with the size of 16 cm × 16 cm which is accomplished by mounting the sensor node using conductive paste on the printed circuit board. Active Cable with redundant structure is printed on the PI substrate to connect each sensor node. The electrodes are also fabricated via the inkjet printing technology and attached beneath the Bio-Patch. The electric connection between the electrode and a sensor node (one input of the AFE) is made via through-hole in the substrate. The use of PI substrate and the integration of the bare die on the PI enable an ultra-thin patch solution with a total thickness less than 1.5 mm. 8 sensor nodes are deployed and mounted on specific points of a printed flexible circuit board layer (corresponding to 8 positions on the subject’s chest). A flat soft battery from Enfucell (0.7 mm thickness, 1.4 g weight) is employed as the power supply. To protect the circuit, a thin plastic foil is applied to cover the patch. The Bio-Patch can be easily attached to a subject’s chest by an elastic bandage or utilizing medical compatible adhesive at the bottom layer to ensure a good contact. Concurrent multi-channel ECG signal measurements are performed using the implemented Bio-Patch prototype. More results are reported in the attached Paper II [32]. 44 4.5 Conclusion In this chapter, we present the implementation work of applying the Bio-electric SoC into a wearable healthcare device. Customized electrodes are fabricated using inkjet printing technology, and their performances are investigated by connecting them to the SoC and examined through in-vivo tests. The reliability of inkjet printed silver traces on flexible substrate is evaluated through the static bending test. The purpose of this work is not limited to compare the printed electrodes with the commercial ones. The main purpose is to present the concept of a hybrid bio-sensing system and to demonstrate a workable prototype using digital inkjet printing technology to facilitate the integration of the customized SoC on a flexible substrate. In addition, we intended to evaluate the possibility and feasibility of integrating the silicon SoC with Printed Electronics. Two prototypes of Bio-Patch are implemented, the first one is based on photo paper substrate and the second one is built on PI substrate. The patches get rid of cumbersome connecting wires and avoid complicated electrodes configuration/placement, increasing the comfort of the user. Preliminary measurement results show that ECG signals are successfully measured when applied on subject’s chest. The photo paper base Bio-Patch has advantages in terms of low cost and disposability after use, while the PI based Bio-Patch is featured with better reliability and concurrent measurement of multi-channel ECG signals. . 45 46 Chapter 5 Summary and Future Works 5.1 Summary This thesis presents and investigates the design and implementation of hybrid sensor devices for enabling wearable pervasive healthcare systems. An Intelligent Electrode has been proposed by integrating a compact SoC into a conductive electrode. A serial connection method has been proposed to interconnect multiple Intelligent Electrodes or sensor nodes in a serial manner to establish a body sensor network. This method can significantly improve user’s comfort by drastically reducing the number of connecting cables. A low-power, low-noise Bio-electric SoC has been designed and fabricated to meet the requirements for weak bio-electrical signal acquisition, as well as the requirements for serial communication and network management. This thesis has focused not only on customized circuit design and implementation on silicon chip, but also on sensor miniaturization and system integration through printing technology on flexible substrates, as well as the corresponding function verification and performance evaluation through in-vivo tests: A fully integrated SoC has been implemented. The SoC has programmable gain and bandwidth to accommodate different bio-electrical signals. Experiment results from in-vivo tests have proven that the fabricated SoC can be used to measure ECG, EMG, EOG, and EEG signal. Also, the tunable gain and bandwidth enable the SoC to interface with customized electrodes with various size and distance to form a novel sensor or sensing system. On-chip memory allows the buffering of the digital bio-signal. The implemented serial communication protocol enables the output of each sensor node as serial data stream, and allows the establishment of a body sensor network. 47 The performance of printed electrodes and interconnections on flexible substrates have been investigated and the results confirm the inkjet printing technology can be successfully employed to print circuit patterns and sensors. By connecting the printed electrodes to the SoC measurement board, high quality ECG signal and EMG signal have been successfully measured. Experimental results show that the printed electrodes are fully compatible with the implemented SoC. It is feasible to print customized electrodes and other patterned electronics on flexible substrates such as photo paper or PI substrate. Static bending tests have been applied to verify the reliability of the printed patterns on flexible substrates. Sliver traces printed on photo paper substrate and PI substrate show good electrical performance. The photo paper substrate is featured with low cost and disposability after use. PI based silver lines exhibit remarkable reliability, which can survive from extreme bending test. Sensor miniaturization and system integration using inkjet printing technology on flexible substrates have been investigated. The design concept of combining compact SoC and printed electronics to form a hybrid-integrated system with the potential to enable wearable healthcare applications has been demonstrated. Two functioning Bio-Patch prototypes have been implemented: one is photo paper based, and the other one is on the PI substrate. Preliminary experiment results have shown that ECG signals are successfully measured through in-vivo tests. 5.2 Future Works Advanced biomedical technology is indispensible from personal and global healthcare and wellbeing since it contributes to improve the life quality, and saving lives in many occasions. As long as there are people suffering from diseases, progresses on biomedical technologies will never stop. Some remarks are added below, which are the key features or function blocks could be added or improved in the future to make it a more advanced system. High resolution, low power ADC: For sophisticated medical device, such as EEG recorder, high resolution ADC is required, for example 12-bit or 14-bit ADC, which can identify quite small changes on the detected signal. Meanwhile, low power is also required to effectively extend the battery life for a portable or wearable medical system. Integration of wireless link: A full integration solution needs a wireless transmission module to transmit the measured bio-signals to a smart phone or base 48 49 station. The integration of wireless transceiver can further reduce the sensor size and make the whole system more compact as well. More sensor interfaces: Future healthcare monitoring systems require multi-sensory bio-signals for assessing the patient health condition. More sensor interfaces are needed to accommodate more external sensors, and therefore achieve the target ‘one-chip multi-sensor’. Explore the developed measuring and system integration technologies in non-clinical area, such as sport and fitness. A frequency-tunable sine wave generator can be designed and implemented on chip to measure electrical bio-impedance for body composition assessment, which is a useful parameter to track in fitness and wellness applications. Display and diagnosis on smart phone: One attractive application is to use a smart phone to receive and display the measured bio-signals and even make a diagnosis in certain cases. This way the user can directly check his/her health condition via the smart phone, which makes the user actively engaged in their own care process, i.e. enabling ‘self-care’. Explore other technologies which could be integrated to form a more comfortable and more efficient wearable health care system, like the use of Smart Textiles and textile-electronic integration technology to form smart garments. 49 50 Bibliography 1. European Commission Eurostat: Causes of death statistics 2011. Available from: http://epp.eurostat.ec.europa.eu/statistics_explained/index.php/Causes_of_deat h_statistics. 2. H.-C. Kung, D.L. Hoyert, J. Xu and Murphy S.L., "Deaths: final data for 2005". National vital statistics reports 56(10). vol: pp. 5-9. Apr. 2008 3. Y. T. Zhang, Y. S. Yan and Poon C.C.Y., "Some Perspectives on Affordable Healthcare Systems in China," pp.6154, , in IEEE 29th Annual International Conference of the Engineering in Medicine and Biology Society, (EMBS 2007). Aug., 2007. 4. Anind K. Dey and Estrin D., "Pervasive healthcare 2010: two perspectives", in IEEE Pervasive Computing. 2011. p. 8-11. 5. N. Saranummi, "IT Applications for Pervasive, Personal, and Personalized Health". IEEE Transactions on Information Technology in Biomedicine. vol 12(1): pp. 1-4. 2008 6. J. Fayn and Rubel P., "Toward a Personal Health Society in Cardiology". IEEE Transactions on Information Technology in Biomedicine. vol 14(2): pp. 401-409. 7. Xiao-Fei Teng, Yuan-Ting Zhang, Poon C. C. Y. and Bonato P., "Wearable Medical Systems for p-Health". Biomedical Engineering, IEEE Reviews in. vol 1: pp. 62-74. 2008 8. Anind K. Dey and Estrin D., "Perspectives on Pervasive Health from Some of the Field's Leading Researchers", in IEEE Pervasive Computing. 2011. p. 4-7. 9. Mulder Ingrid, Schikhof Yvonne, Vastenburg Martijn, Card Alan, Dunn Tory, Komninos Andreas, Mcgee-Lennon Marilyn, Santcroos Mark, Tiotto Gabriele, Van Gils Mieke, Van 'T Klooster Jan-Willem, Veys Annelies, and Mohammed Zarifi Eslami, "Designing with Care: The Future of Pervasive Healthcare". IEEE Pervasive Computing. vol 8(4): pp. 85-88. 2009 10. V. Stanford, "Using pervasive computing to deliver elder care". IEEE Pervasive Computing. vol 1(1): pp. 10-13. 2002 11. P. Langendoerfer, F. Vater and Piotrowski K. "Customizing sensor nodes and software for individual pervasive health applications". in Pervasive Computing Technologies for Healthcare, 2009. PervasiveHealth 2009. 3rd 51 International Conference on. 2009. 52 12. B. Venkatalakshmi, A. P. Renold and Packiam R. S. L. "Smart RFID care [SRC] for pervasive health care system". in Communication Software and Networks (ICCSN), 2011 IEEE 3rd International Conference on. 13. Kun-Chieh Yeh, Y. C. Chen, C. C. Chen and Chen Reuy-Shun. "An RFID-Based Software Agent Framework on Pervasive Health Service". in Biomedical Engineering and Informatics, 2009. BMEI '09. 2nd International Conference on. 2009. 14. C. Serodio, P. Mestre, L. M. Joao and Carlos A. C. C. "Pervasive WSN based solutions applied to Health and Life-support systems". in IECON 2010 - 36th Annual Conference on IEEE Industrial Electronics Society. 15. G. N. Priya, P. Anandhakumar and Maheswari K. G. "Dynamic scheduler - a pervasive healthcare system in smart hospitals using RFID". in Computing, Communication and Networking, 2008. ICCCn 2008. International Conference on. 2008. 16. P. Crilly and Muthukkumarasamy V. "Using smart phones and body sensors to deliver pervasive mobile personal healthcare". in Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP), 2010 Sixth International Conference on. 17. L. M. Borges, N. Barroca, F. J. Velez and Lebres A. S. "Smart-clothing wireless flex sensor belt network for foetal health monitoring". in Pervasive Computing Technologies for Healthcare, 2009. PervasiveHealth 2009. 3rd International Conference on. 2009. 18. A. Totter, D. Bonaldi and Majoe D. "A human-centered approach to the design and evaluation of wearable sensors - Framework and case study". in Pervasive Computing and Applications (ICPCA), 2011 6th International Conference on. 19. M. Harris and Habetha J. "The MyHeart project: A framework for personal health care applications". in Computers in Cardiology, 2007. 2007. 20. H. Reiter, J. Muehlsteff and Sipila A. "Medical application and clinical validation for reliable and trustworthy physiological monitoring using functional textiles: Experience from the HeartCycle and MyHeart project". in Engineering in Medicine and Biology Society,EMBC, 2011 Annual International Conference of the IEEE. 21. H. Reiter. "HeartCycle: Beyond building demonstrators. A structured approach to develop, implement and validate healthcare innovations in telemonitoring". in Engineering in Medicine and Biology Society (EMBC), 2010 Annual International Conference of the IEEE. 22. G. Yang, J. Chen, Y. Cao, H. Tenhunen and Zheng L.-R. "A novel wearable ECG monitoring system based on active-cable and intelligent electrodes". in e-health Networking, Applications and Services, 2008. HealthCom 2008. 10th International Conference on. 2008. 53 23. G. Yang, Y. Cao, J. Chen, H. Tenhunen and Zheng L.-R. "An Active-Cable Connected ECG Monitoring System for Ubiquitous Healthcare". in Convergence and Hybrid Information Technology, 2008. ICCIT '08. Third International Conference on. 2008. 24. G. Yang, J. Chen, H. Tenhunen and Zheng L.-R. "An ASIC Solution for Intelligent Electrodes and Active-cable Used in a Wearable ECG Monitoring System". in International Conference on Biomedical Electronics and Devices, 2009. BIODEVICES '09. . 2009. 25. G. Yang, J. Chen, F. Jonsson, H. Tenhunen and Zheng L.-R. "A 1.0 V 78 uW reconfigurable ASIC embedded in an intelligent electrode for continuous remote ECG applications". in Engineering in Medicine and Biology Society, 2009. EMBC 2009. Annual International Conference of the IEEE. 2009. 26. G. Yang, J. Mao, H. Tenhunen and Zheng L.-R. "Design of a self-organized Intelligent Electrode for synchronous measurement of multiple bio-signals in a wearable healthcare monitoring system". in Applied Sciences in Biomedical and Communication Technologies (ISABEL'10), 2010 3rd International Symposium on. 2010. 27. G. Yang, Q. Wan and Zheng L.-R. "Bio-Chip ASIC and Printed Flexible Cable on Paper Substrate for Wearable Healthcare Applications". in Applied Sciences in Biomedical and Communication Technologies (ISABEL'11), 2011 4th International Symposium on. 2011. 28. Q. Wan, G. Yang, Q. Chen and Zheng L.-R. "Electrical performance of inkjet printed flexible cable for ECG monitoring". in 2011 IEEE 20th Conference on Electrical Performance of Electronic Packaging and Systems (EPEPS'11) 2011. 29. G. Yang, J. Chen, F. Jonsson, H. Tenhunen and Zheng L.-R. "A multi-parameter bio-electric ASIC sensor with integrated 2-wire data transmission protocol for wearable healthcare system". in Design, Automation & Test in Europe Conference & Exhibition (DATE'12), 2012. 30. L. Xie, G. Yang, M. Mantysalo, F. Jonsson and Zheng L.-R. "A System-on-Chip and Paper-based Inkjet Printed Electrodes for a Hybrid Wearable Bio-Sensing System". in Engineering in Medicine and Biology Society, 2012. EMBC 2012. Annual International Conference of the IEEE. 2012. 31. G. Yang, L. Xie, M. Mantysalo, J. Chen, H. Tenhunen, and Zheng L.-R., "Bio-Patch Design and Implementation Based on a Low-Power System-on-Chip and Paper-based Inkjet Printing Technology". IEEE Transactions on Information Technology in Biomedicine (TITB). vol 16(6): pp. 1043-1050. 2012 32. G. Yang, J. Chen, L. Xie, J. Mao, H. Tenhunen, and Zheng L.-R., "A Hybrid Low Power Bio-Patch for Body Surface Potential Measurement". IEEE Journal of Biomecial and Health Informatics (J-BHI) (accept). vol. 2013 53 54 33. J. Chen, L. Rong, F. Jonsson, G. Yang and Zheng L-R., "The Design of All-Digital Polar Transmitter Based on ADPLL and Phase Synchronized Sigma Delta Modulator". IEEE Journal of Solid-State Circuits. vol 47(5): pp. 1154-1164. 2012 34. M. Baghaei-Nejad, D. S. Mendoza, Z. Zhuo, S. Radiom, G. Gielen, L.-R. Zheng, and Tenhunen H. A remote-powered RFID tag with 10Mb/s UWB uplink and −18.5dBm sensitivity UHF downlink in 0.18µm CMOS. in Solid-State Circuits Conference - Digest of Technical Papers, 2009. ISSCC 2009. IEEE International. 35. Semiconductor Nordic, http://www.nordicsemi.com/eng/Products/Bluetooth-R-low-energy/nRF8001. 2013. 36. R. Plonsey, Bioelectric Phenomena. 1969, New York: McGraw-Hill. 37. L. A. Geddes and Baker L. E., Principles of Applied Biomedical Instrumentation. 1989, New York: Wiley. 38. A. Lymberis. "Smart wearable systems for personalised health management: current R&D and future challenges,". in Engineering in Medicine and Biology Society, 2003. Proceedings of the 25th Annual International Conference of the IEEE. 39. X. D. Zou, X. Xu, L. Yao and Lian Y., "A 1-V 450-nW Fully Integrated Programmable Biomedical Sensor Interface Chip". IEEE Journal of Solid-State Circuits. vol 44(4): pp. 1067-1077. 2009 40. N.V. Thakor, “Biopotentials and Electrophysiology Measurement,”in The Measurement, Instrumentation and Sensors Handbook 2000, Madison: CRC Press LLC. 41. E. Braunwald, Heart Disease: A Textbook of Cardiovascular Medicine. 1997, Philadelphia: W.B. Saunders Co.,. 108. 42. A. S. Gevins and Aminoff M. J., "Electroencephalography: brain electrical activity," in Encyclopedia of Medical Devices and Instrumentation. 1988, New York: Wiley. 1084-1107. 43. E. Niedermeyer and Silva F. L. Da, Electroencephalography. Baltimore: Urban & Schwarzenberg. 1987. 44. M. Brown, M. Marmor, Vaegan, E. Zrenner, M.Brigell, and Bach M., ISCEV Standard for Clinical Electro-oculography (EOG) (2006). 2006. 205-212. 45. A. Bulling, J. A. Ward, Gellersen H. and Tröster G. "Robust Recognition of Reading Activity in Transit Using Wearable Electrooculography,". in the 6th International Conference on Pervasive Computing (Pervasive 2008). 2008: Springer. 46. A. Bulling, D. Roggen and Tröster G., "Wearable EOG goggles: Seamless sensing and context-awareness in everyday environments,". Journal of Ambient Intelligence and Smart Environments. vol 1: pp. 157-171. 2009 47. S. Micera, T. Keller, M. Lawrence, M. Morari and Popovic D. B., "Wearable 55 Neural Prostheses". IEEE Engineering in Medicine and Biology Magazine. vol 29(3): pp. 64-69. 2010 48. A. Searle and Kirkup L., "A direct comparison of wet, dry and insulating bioelectric recording electrodes". Physiological Measurement vol 21(2): pp. 271-283. May, 2000 49. E. Huigen, A. Peper and C. A. Grimbergen, "Investigation into the origin of the noise of surface electrodes". Med. Biological Eng. Comput. vol 40(3): pp. 332-338. 2002 50. M. Spach, R. Barr, J. Havstad and Long E., “Skin-electrode impedance and its effect on recording cardiac potentials,”. Circulation. vol 34: pp. 649-656. 1966 51. M. Griffith, W. Portnoy, L. Stotts and Day J., "Improved capacitive electrocardiogram electrodes for burn applications,". Med. Biol. Eng. Comput. vol 17(5): pp. 641-646. 1979 52. S. Bouwstra, L. Feijs, C. Wei and Oetomo S. B. "Smart Jacket Design for Neonatal Monitoring with Wearable Sensors,". in Wearable and Implantable Body Sensor Networks, 2009. BSN 2009. Sixth International Workshop on. 53. T. Kato, A. Ueno, S. Kataoka, H. Hoshino and Ishiyama Y. "An Application of Capacitive Electrode for Detecting Electrocardiogram of Neonates and Infants,". in Engineering in Medicine and Biology Society, 2006. EMBS '06. 28th Annual International Conference of the IEEE. 54. T. J. Sullivan, S. R. Deiss, T.-P. Jung and Cauwenberghs G. "A brain-machine interface using dry-contact, low-noise EEG sensors". in Circuits and Systems, 2008. ISCAS 2008. IEEE International Symposium on. 55. C. R. Merritt, H. T. Nagle and Grant E., "Fabric-Based Active Electrode Design and Fabrication for Health Monitoring Clothing". Information Technology in Biomedicine, IEEE Transactions on. vol 13(2): pp. 274-280. 2009 56. Gilsoo Cho, Keesam Jeong, Min Joo Paik, Youngeun Kwun and Moonsoo Sung, "Performance Evaluation of Textile-Based Electrodes and Motion Sensors for Smart Clothing". Sensors Journal, IEEE. vol 11(12): pp. 3183-3193. 57. T. Degen, S. Torrent and Jackel H., "Low-Noise Two-Wired Buffer Electrodes for Bioelectric Amplifiers,". Biomedical Engineering, IEEE Transactions on. vol 54(7): pp. 1328-1332. 2007 58. T. J. Sullivan, S. R. Deiss and Cauwenberghs G. "A Low-Noise, Non-Contact EEG/ECG Sensor". in Biomedical Circuits and Systems Conference, 2007. BIOCAS 2007. IEEE. 59. Y. M. Chi, S. R. Deiss and Cauwenberghs G. "Non-contact Low Power EEG/ECG Electrode for High Density Wearable Biopotential Sensor Networks". in Wearable and Implantable Body Sensor Networks, 2009. BSN 2009. Sixth International Workshop on. 55 56 60. Y. M. Chi and Cauwenberghs G. "Micropower integrated bioamplifier and auto-ranging ADC for wireless and implantable medical instrumentation". in ESSCIRC, 2010 Proceedings of the. 2010. 61. Yong Gyu Lim, Ko Keun Kim and Suk Park, "ECG measurement on a chair without conductive contact,". Biomedical Engineering, IEEE Transactions on. vol 53(5): pp. 956-959. 2006 62. Yong Gyu Lim, Ko Keun Kim and Kwang Suk Park, "ECG Recording on a Bed During Sleep Without Direct Skin-Contact,". Biomedical Engineering, IEEE Transactions on. vol 54(4): pp. 718-725. 2007 63. J. Yoo, L. Yan, S. Lee, H. Kim and Yoo H.-J., "A Wearable ECG Acquisition System With Compact Planar-Fashionable Circuit Board-Based Shirt". IEEE Transactions on Information Technology in Biomedicine. vol 13(6): pp. 897-902. 2009 64. Shuenn-Yuh Lee, Su M. Y., Ming-Chun Liang, You-Yin Chen, Cheng-Han Hsieh, Chung-Min Yang, Hsin-Yi Lai, Jou-Wei Lin, and Qiang Fang, A Programmable Implantable Microstimulator SoC With Wireless Telemetry: Application in Closed-Loop Endocardial Stimulation for Cardiac Pacemaker. Biomedical Circuits and Systems, IEEE Transactions on. vol 5(6): pp. 511-522. 65. M. Vidojkovic M., X. Huang X., P. Harpe P., S. Rampu S., C. Zhou C., L. Huang, J. Van De Molengraft, K. Imamura, B. Busze, F. Bouwens, M. Konijnenburg, J. Santana, A. Breeschoten, J. Huisken, K. Philips, G. Dolmans, and Groot H. De, A 2.4 GHz ULP OOK Single-Chip Transceiver for Healthcare Applications. Biomedical Circuits and Systems, IEEE Transactions on. vol 5(6): pp. 523-534. 66. M. T. Salam, M. Sawan and Khoa N. Dang, "A Novel Low-Power-Implantable Epileptic Seizure-Onset Detector". Biomedical Circuits and Systems, IEEE Transactions on. vol 5(6): pp. 568-578. 67. W. Wattanapanitch and Sarpeshkar R., "A Low-Power 32-Channel Digitally Programmable Neural Recording Integrated Circuit". Biomedical Circuits and Systems, IEEE Transactions on. vol 5(6): pp. 592-602. 68. M. J. Burke and Gleeson D. T., "A micropower dry-electrode ECG preamplifier". Biomedical Engineering, IEEE Transactions on. vol 47(2): pp. 155-162. 2000 69. Hui Zhang, Yajie Qin and Zhiliang Hong. "A 1.8-V 770-nW biopotential acquistion system for portable applications". in Biomedical Circuits and Systems Conference, 2009. BioCAS 2009. IEEE. 2009. 70. Szu-Chieh Liu and Kea-Tiong Tang. "A low-voltage low-power sigma-delta modulator for bio-potential signals". in Life Science Systems and Applications Workshop (LiSSA), 2011 IEEE/NIH. 2011. 71. Xiaoyuan Xu, Xiaodan Zou, Libin Yao and Yong Lian. "A 1-V 450-nW fully integrated biomedical sensor interface system". in VLSI Circuits, 2008 IEEE 57 Symposium on. 2008. 72. W. Massagram, N. Hafner, M. Chen, L. Macchiarulo, V. M. Lubecke, and Boric-Lubecke O., "Digital Heart-Rate Variability Parameter Monitoring and Assessment ASIC". IEEE Transactions on Biomedical Circuits and Systems. vol 4(1): pp. 19-26. 2010 73. B. D. Farnsworth, D. M. Talyor, R. J. Triolo and D. J. Young. "Wireless in vivo EMG sensor for intelligent prosthetic control". in Solid-State Sensors, Actuators and Microsystems Conference, 2009. TRANSDUCERS 2009. International. 74. R. F. Yazicioglu, P. Merken, R. Puers and C. Van Hoof, "A 60 uW 60 nV/vHz Readout Front-End for Portable Biopotential Acquisition Systems". Solid-State Circuits, IEEE Journal of. vol 42(5): pp. 1100-1110. 2007 75. N. Verma, A. Shoeb, J. Bohorquez, J. Dawson, J. Guttag, and Chandrakasan A. P., "A Micro-Power EEG Acquisition SoC With Integrated Feature Extraction Processor for a Chronic Seizure Detection System". IEEE Journal of Solid-State Circuits. vol 45(4): pp. 804-816. 2010 76. Mollazadeh Mohsen, Murari Kartikeya, Cauwenberghs Gert and Thakor Nitish. "From spikes to EEG: Integrated multichannel and selective acquisition of neuropotentials". in Engineering in Medicine and Biology Society, 2008. EMBS 2008. 30th Annual International Conference of the IEEE. 2008. 77. Sheung Wai Fung, Bing Liu, Jie Yuan and Qing Guo. "A low-noise monolithic CMOS bio-potential detector". in Circuits and Systems, 2009. ISCAS 2009. IEEE International Symposium on. 2009. 78. M. Mollazadeh, K. Murari, G. Cauwenberghs and Thakor N., "Micropower CMOS Integrated Low-Noise Amplification, Filtering, and Digitization of Multimodal Neuropotentials". IEEE Transactions on Biomedical Circuits and Systems. vol 3(1): pp. 1-10. 2009 79. R. R. Harrison and Charles C., "A low-power low-noise CMOS amplifier for neural recording applications". Solid-State Circuits, IEEE Journal of. vol 38(6): pp. 958-965. 2003 80. F. Shahrokhi, K. Abdelhalim, D. Serletis, P. L. Carlen and Genov R., "The 128-Channel Fully Differential Digital Integrated Neural Recording and Stimulation Interface". IEEE Transactions on Biomedical Circuits and Systems. vol 4(3): pp. 149-161. 2010 81. B. Gosselin, M. Sawan and Kerherve E., "Linear-Phase Delay Filters for Ultra-Low-Power Signal Processing in Neural Recording Implants". IEEE Transactions on Biomedical Circuits and Systems. vol 4(3): pp. 171-180. 2010 82. H. Miranda, V. Gilja, C. A. Chestek, K. V. Shenoy and Meng T. H., "HermesD: A High-Rate Long-Range Wireless Transmission System for Simultaneous Multichannel Neural Recording Applications". IEEE Transactions on Biomedical Circuits and Systems. vol 4(3): pp. 181-191. 2010 83. Seung Bae Lee, Hyung-Min Lee, Kiani M., Uei-Ming Jow and Ghovanloo M. 57 "An inductively powered scalable 32-channel wireless neural recording system-on-a-chip for neuroscience applications". in Solid-State Circuits Conference Digest of Technical Papers (ISSCC), 2010 IEEE International. 2010. 58 84. Moosung Chae, Wentai Liu, Zhi Yang, Tungchien Chen, Jungsuk Kim, Sivaprakasam M., and Yuce M. "A 128-Channel 6mW Wireless Neural Recording IC with On-the-Fly Spike Sorting and UWB Tansmitter". in Solid-State Circuits Conference, 2008. ISSCC 2008. Digest of Technical Papers. IEEE International. 2008. 85. Horiuchi T., Swindell T., Sander D. and Abshier P. "A low-power CMOS neural amplifier with amplitude measurements for spike sorting". in Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on. 2004. 86. W. Wattanapanitch, M. Fee and Sarpeshkar R., "An Energy-Efficient Micropower Neural Recording Amplifier". Biomedical Circuits and Systems, IEEE Transactions on. vol 1(2): pp. 136-147. 2007 87. S. K. Kelly and Wyatt J. L., "A Power-Efficient Neural Tissue Stimulator With Energy Recovery". Biomedical Circuits and Systems, IEEE Transactions on. vol 5(1): pp. 20-29. 88. K. Sooksood, T. Stieglitz and Ortmanns M., "An Active Approach for Charge Balancing in Functional Electrical Stimulation". IEEE Transactions on Biomedical Circuits and Systems. vol 4(3): pp. 162-170. 2010 89. D. Loi, C. Carboni, G. Angius, G. N. Angotzi, M. Barbaro, L. Raffo, S. Raspopovic, and Navarro X., "Peripheral Neural Activity Recording and Stimulation System". IEEE Transactions on Biomedical Circuits and Systems. vol 5(4): pp. 368-379. 2011 90. P. J. Langlois, A. Demosthenous, I. Pachnis and Donaldson N., "High-Power Integrated Stimulator Output Stages With Floating Discharge Over a Wide Voltage Range for Nerve Stimulation". IEEE Transactions on Biomedical Circuits and Systems. vol 4(1): pp. 39-48. 2010 91. S. K. Kelly and Wyatt J. L., "A Power-Efficient Neural Tissue Stimulator With Energy Recovery". IEEE Transactions on Biomedical Circuits and Systems. vol 5(1): pp. 20-29. 2011 92. S. Ethier and M. Sawan, "Exponential Current Pulse Generation for Efficient Very High-Impedance Multisite Stimulation". IEEE Transactions on Biomedical Circuits and Systems. vol 5(1): pp. 30-38. 2011 93. P. Livi, F. Heer, U. Frey, D. J. Bakkum and Hierlemann A., "Compact Voltage and Current Stimulation Buffer for High-Density Microelectrode Arrays". IEEE Transactions on Biomedical Circuits and Systems. vol 4(6): pp. 372-378. 2010 94. J. H. Nagle, “Biopotential amplifiers”, in The Biomedical Engineering Handbook. 1995: CRC Press. 1185-1195. 59 95. H. P. Le, J. Singh, L. Hiremath, V. Mallapur and A. Stojcevski, "Ultra-low-power variable-resolution successive approximation ADC for biomedical application". Electronics Letters. vol 41(11): pp. 634-635. 2005 96. Robert Pierre-Yves, Gosselin Benoit, Ayoub Amer Elias and Sawan Mohamad. "An Ultra-Low-Power Successive-Approximation-Based ADC for Implantable Sensing Devices". in Circuits and Systems, 2006. MWSCAS '06. 49th IEEE International Midwest Symposium on. 2006. 97. Y. Susanti, P. K. Chan and Ong V. "An ultra low-power Successive Approximation ADC using an offset-biased auto-zero comparator". in Circuits and Systems, 2008. APCCAS 2008. IEEE Asia Pacific Conference on. 2008. 98. Arbat A., Dieguez A. and Samitier J. "An Ultra Low Power Successive Approximation ADC with Selectable Resolution in 0.13 μm CMOS Technology". in Electronics, Circuits and Systems, 2006. ICECS '06. 13th IEEE International Conference on. 2006. 99. P. Kamalinejad, S. Mirabbasi and Leung V. C. M. "An ultra-low-power SAR ADC with an area-efficient DAC architecture". in 2011 IEEE International Symposium on Circuits and Systems (ISCAS). 2011. 100. T. G. Rabuske, F. A. Rabuske and Rodrigues C. R. "A novel energy efficient digital controller for charge sharing successive approximation ADC". in 2011 IEEE Second Latin American Symposium on Circuits and Systems (LASCAS). 2011. 101. J. C. Huhta and J. G. Webster, "60-Hz Interference in Electrocardiography,". Biomedical Engineering, IEEE Transactions on. vol BME-20(2): pp. 91-101. 1973 102. C. J. D. Luca (2002) "Surface Electromyography: Detection and Recording " by DelSys Incorporated. 103. Noureddin B., Lawrence P. D. and Birch G. E. "Time-frequency Analysis of Eye Blinks and Saccades in EOG for EEG Artifact Removal". in Neural Engineering, 2007. CNE '07. 3rd International IEEE/EMBS Conference on. 2007. 104. Mantysalo M., Pekkanen V., Kaija K., Niittynen J., Koskinen S., Halonen E., Mansikkamaki P., and Hameenoja O. "Capability of inkjet technology in electronics manufacturing". in Electronic Components and Technology Conference, 2009. ECTC 2009. 59th. 2009. 105. M. Singh , H. M. Haverinen, P. Dhagat and Jabbour G. E., "Inkjet Printing-Process and Its Application". Adv. Mater. vol 22: pp. 673–685. 2009 106. V. Pekkanen, M. Mäntysalo, K. Kaija, P. Mansikkamäki and Kunnari E., "Utilizing inkjet printing to fabricate electrical interconnections in a System-in-Package". ELSEVIER Microelectronic Engineering. vol 87(11): pp. 2382-2390. 2010 107. M. Mantysalo, L. Xie, F. Jonsson, Y. Feng, A. L. Cabezas, and Zheng L-R. "System integration of smart packages using printed electronics". in 59 Electronic Components and Technology Conference (ECTC), 2012 IEEE 62nd. 2012. 60 108. A. Sridhar, T. Blaudeck and Baumann R. R., "Inkjet Printing as a Key Enabling Technology for Printed Electronics". Material Matters. vol 6. 2011 109. A. Casson, D. Yates, S. Smith, J. Duncan and Rodriguez-Villegas E., "Wearable Electroencephalography". IEEE Engineering in Medicine and Biology Magazine. vol 29(3): pp. 44-56. 2010 110. V. Pekkanen, M. Mäntysalo, K. Kaija, P. Mansikkamäki, E. Kunnari, K. Laine, J. Niittynen, S. Koskinen, E. Halonen, and Caglar U., "Utilizing inkjet printing to fabricate electrical interconnections in a System-in-Package". ELSEVIER Microelectronic Engineering. vol 87(11): pp. 2382-2390. 2010 111. Bing An, Xiong-Hui Cai, Hua-Bin Chu, Xiao-Wei Lai, Feng-Shun Wu, and Yi-Ping Wu. "Flex Reliability of RFID Inlays Assembled by Anisotropic Conductive Adhesive". in High Density packaging and Microsystem Integration, 2007. HDP '07. International Symposium on. 2007. 112. Xiong-Hui Cai, Bing An, Yi-Ping Wu, Feng-Shun Wu and Xiao-Wei Lai. "Research on the contact resistance and reliability of flexible RFID tag inlays packaged by anisotropic conductive paste". in Electronic Packaging Technology & High Density Packaging, 2008. ICEPT-HDP 2008. International Conference on. 2008. 113. J. S. Rasul, "Chip on paper technology utilizing anisotropically conductive adhesive for smart label applications" Microelectronics Reliability. vol 44(1): pp. 135-140. 2004 114. Y. Li and C. P. Wong, "Recent advances of conductive adhesives as a lead-free alternative in electronic packaging: Materials, processing, reliability and applications". Elsevier, Materials Science and Engineering. vol 51: pp. 1-35. 2006 115. Su-Tsai Lu and Wen-Hwa Chen, "Reliability and Flexibility of Ultra-Thin Chip-on-Flex (UTCOF) Interconnects With Anisotropic Conductive Adhesive (ACA) Joints". Advanced Packaging, IEEE Transactions on. vol 33(3): pp. 702-712. 2010