Jerry Tsay

A real-time fall detection system using a wearable gait analysis sensor and a Support Vector Machine (SVM) classifier

In this study, we report a custom designed wireless gait analysis sensor (WGAS) system for real-time fall detection using a Support Vector Machine (SVM) classifier. Our WGAS includes a tri-axial accelerometer, 2 gyroscopes and a MSP430 micro-controller. It was worn by the subjects at either the T4 or at the waist level for various intentional falls, Activities of Daily Living (ADL) and the Dynamic Gait Index (DGI) test. The raw data of tri-axial acceleration and angular velocity is wirelessly transmitted from the WGAS to a nearby PC, and then 6 features were extracted for fall classification using a SVM (Support Vector Machine) classifier. We achieved 98.8% and 98.7% fall classification accuracies from the data at the T4 and belt positions, respectively. Moreover, the preliminary data demonstrates an impressive overall specificity of 99.5% and an overall sensitivity of 97.0% for this WGAS real-time fall detection system.

Robust phased array non-contact vital signs monitoring in an office cubicle setting

Doppler-based non-contact vital signs (NCVS) sensor systems can monitor the heart and respiration rates without touching the patient, but the accuracy of the NCVS sensing can be degraded considerably by background clutters and movements artifacts from the monitored subject or other motions in the measurement environment. We have, therefore, developed a high directivity phased-array antenna NCVS system that can significantly increase the sensing accuracy and the monitoring range in a typical office cubicle setting. Depending on the position of the beam spot illuminated on the chest, our custom-made beam-steerable phased array NCVS system at 2.4GHz has demonstrated the accuracy of the heart rate monitoring to be within +/- 2 bpm (beat-per-minute) for 92.5%-99.2% of the time at 0.75m. As the monitoring range is extended to 1-1.5m with beam directly illuminating on the center of the chest, the heart rate monitoring accuracy remains within +/- 5 bpm above 78% of time. With more improvement, this phased array NCVS system should benefit patients considerably for telemedicine and remote monitoring, as the patients vital signs can be monitored comfortably and continuously in both home and office settings, eventually capable of detecting potential sleep apnea and other respiratory or cardiovascular symptoms in real-time.

A phased array non-contact vital signs sensor with automatic beam steering

Doppler-based non-contact vital signs (NCVS) sensor systems have the ability to monitor heart and respiration rates of patients without physical contacts. Because the accuracy of a NCVS sensor can deteriorate quickly in a noisy or cluttered environment, and that patients confined on their beds have different physical sizes and microwave signatures and will still move naturally (though not frequently), continuous NCVS monitoring that can work well for all individuals is very difficult. Therefore, we have developed a highly directive phased-array antenna NCVS system that can perform automatic electronic beam steering for continuous NCVS monitoring with considerably improved monitoring accuracy over that obtained from the Doppler radar with a fixed beam. Our NCVS system includes an automatic beam steering algorithm, and has achieved heart rate measurement accuracy of nearly 95% within 5 beat-per-minute (BPM) vs. reference at our engineering lab.

Accurate and continuous non-contact vital signs monitoring using phased array antennas in a clutter-free anechoic chamber

Continuous and accurate monitoring of human vital signs is an important part of the healthcare industry, as it is the basic means by which the clinicians can determine the instantaneous status of their patients. Doppler-based noncontact vital signs (NCVS) sensor systems can monitor the heart and respiration rates without touching the patient, but it has been observed that that the accuracy of these NCVS sensors can be diminished by reflections from background clutters in the measurement environment, and that high directivity antennas can increase the sensing accuracy. Therefore, this work explores a NCVS sensor with continuous data taken inside an anechoic chamber where the background cluttering is negligible. In addition, a high directivity custom-made beam-steerable phased array antenna system is used to improve the performance and functionality of the 2.4GHz NCVS sensor we have built. We believe this work is the 1st systematic study using Doppler-based phased array systems for NCVS sensing performed in a clutter-free anechoic chamber.

High-efficiency silicon RF power amplifier design – current status and future outlook

Silicon RF power amplifiers (PAs) are in various RF front end modules (FEMs) today for handset and WLAN applications. Even though III-V semiconductor-based RF PAs can still offer superior frequency and breakdown performance with higher Pout and power-added-efficiency (PAE) and faster time-to-market, silicon-based RF PAs do have the advantages in offering higher monolithic integration with added functionalities (e.g., on-chip digital control and selection on power level, modulation, frequency band, matching, predistortion, etc.), which can translate to lower cost and smaller sizes attractive for broadband multi-mode multi-band handset transmitters. Therefore, some key techniques for designing high-efficiency 4G/5G/WLAN broadband wireless silicon PAs will be discussed.

A real-time robust fall detection system using a wireless gait analysis sensor and an Artificial Neural Network

This paper describes our custom-designed wireless gait analysis sensor (WGAS) system developed and tested for real-time fall detection. The WGAS is capable of differentiating falls vs. Activities of Daily Living (ADL) and the Dynamic Gait Index (DGI) performed by young volunteers using a Back Propagation Artificial Neural Network (BP ANN) algorithm. The WGAS, which includes a tri-axial accelerometer, 2 gyroscopes, and a MSP430 microcontroller is worn by the subjects at either T4 (at back) or the belt-clip positions (in front of the waist) for the various falls, ADL, and Dynamic Gait Index (DGI) tests. The raw data is wirelessly transmitted from the WGAS to a nearby PC for real-time fall classification, where six features were extracted for the BP ANN. Overall fall classification accuracies of 97.0% and 97.4% have been achieved for the data taken at the T4 and at the belt position, respectively. The preliminary data demonstrates an overall sensitivity of 97.0% and overall specificity of 97.2% for this WGAS fall detection system, showing good promise as a real-time low-cost and effective fall detection device for wireless acute care and wireless assisted living.

Recent progress on high-efficiency CMOS and SiGe RF power amplifier design

The majority of the worlds RF wireless power amplifiers (PA) products are still designed in III-V semiconductors today. However, by taking advantage of nm silicon devices and novel RF system-on-a-chip (SoC) design techniques, several groups have recently reported highly-competitive silicon RF PAs in both CMOS and SiGe BiCMOS technologies with performance rivaling those of the III-V RF PAs. We will, therefore, present an up-to-date review on recent design trends of silicon-based PAs, with the focus on high-efficient broadband wireless and 5G PA design.

A differential SiGe power amplifier using through-silicon-via and envelope-tracking for broadband wireless applications

In this paper, a differential SiGe power amplifier (PA) is designed using a bipolar differential pair in a 0.35-μm SiGe BiCMOS technology with through-silicon-via (TSV). Measured using continuous wave (CW) and Long Term Evolution (LTE) modulated waveforms, significant gain expansion (3-5 dB) is observed. The PA reaches power-added-efficiency (PAE) of 61.7% / 51.2% / 40.0% at Pout = 25.6 / 25.4 / 25.7 dBm at supply voltages of Vcc = 2.8V / 3.3V / 4.2V, respectively, with 800 MHz CW input. With the help of the envelope-tracking (ET) technique, the measured PAE improves by 7.3% / 10.4% / 15.4% compared to the fixed supply PA at power back-off regions at Pout = 19.9 / 22.1 / 22.4 dBm, achieving PAE of 38.4% / 43.4% / 38.6% at 800 MHz for LTE 16QAM 5 MHz and passing the LTE spectrum emission mask (SEM) without predistortion. This SiGe ET-PA shows promise for operation as the medium power (MP) PA for efficiency enhancement in the back-off regions.

Long-term vital sign measurement using a non-contact vital sign sensor inside an office cubicle setting

Heart and respiration rates can be wirelessly measured by extracting the phase shift caused by the periodic displacement of a patients chest wall. We have developed a phased-array Doppler-based non-contact vital sign (NCVS) sensor capable of long-term vital signs monitoring using an automatic patient tracking and movement detection algorithm. Our NCVS sensor achieves non-contact heart rate monitoring with accuracies of over 90% (i.e, within ±5 Beats-Per-Minute vs. a reference sensor) across a large number of data points collected over various days of the week inside a typical office cubicle setting at a distance of 1.5 meters.

Real-Time Classification of Patients with Balance Disorders vs. Normal Subjects Using a Low-Cost Small Wireless Wearable Gait Sensor

Gait analysis using wearable wireless sensors can be an economical, convenient and effective way to provide diagnostic and clinical information for various health-related issues. In this work, our custom designed low-cost wireless gait analysis sensor that contains a basic inertial measurement unit (IMU) was used to collect the gait data for four patients diagnosed with balance disorders and additionally three normal subjects, each performing the Dynamic Gait Index (DGI) tests while wearing the custom wireless gait analysis sensor (WGAS). The small WGAS includes a tri-axial accelerometer integrated circuit (IC), two gyroscopes ICs and a Texas Instruments (TI) MSP430 microcontroller and is worn by each subject at the T4 position during the DGI tests. The raw gait data are wirelessly transmitted from the WGAS to a near-by PC for real-time gait data collection and analysis. In order to perform successful classification of patients vs. normal subjects, we used several different classification algorithms, such as the back propagation artificial neural network (BP-ANN), support vector machine (SVM), k-nearest neighbors (KNN) and binary decision trees (BDT), based on features extracted from the raw gait data of the gyroscopes and accelerometers. When the range was used as the input feature, the overall classification accuracy obtained is 100% with BP-ANN, 98% with SVM, 96% with KNN and 94% using BDT. Similar high classification accuracy results were also achieved when the standard deviation or other values were used as input features to these classifiers. These results show that gait data collected from our very low-cost wearable wireless gait sensor can effectively differentiate patients with balance disorders from normal subjects in real time using various classifiers, the success of which may eventually lead to accurate and objective diagnosis of abnormal human gaits and their underlying etiologies in the future, as more patient data are being collected.