Gallium oxide has become a better alternative to wide-band materials such as GaN, SiC, etc. for radio frequency and power electronics applications. It can achieve higher breakdown voltage in the range of 8MV/cm and ensure greater efficiency compared to GaN and SiC due to its much higher bandgap and Baliga’s figure of merit.  Lateral and enhancement mode devices based on β-Ga2O3 material are one of the cutting-edge topics in the field of power electronic devices.  However, lateral devices are not suitable for applications that require breakdown voltages above 800V and cause significant drift resistance at reverse voltages of more than 200V.  Vertical device structures are used to meet these limitations that offer a wider range of potential applications. Still, the low thermal conductivity of β-Ga2O3 is a major concern which leads to extreme self-heating, that deteriorates the device performance during high voltage operation.  Our research focuses on developing an Electro-Thermal co design of vertical β-Ga2O3 power device considering self-heating effect, incomplete ionization of dopants and temperature-dependent mobility.

This project focuses on the development of a wearable biomedical system capable of forecasting the future of blood glucose levels for Type I diabetic patients. The system takes current blood glucose level, carbohydrate intake, activity signals, and other physiological signals as input and uses a deep learning model to predict the future glucose values at different prediction horizons. Additionally, the parameters from the trained models are deployed into edge devices to make the learned, data-driven, and automatic predictions of future glucose values to help the patient take precautionary measures to avoid hyper/hypoglycemic events.

This project emphasizes on the development of a biomedical system design that helps detect and classify diseases of Neo-natal infants in NICU (Neo-Natal Intensive Care Unit) and respiratory disorders of patients admitted to sleep laboratories. Through active collaboration with the health informatics department, we are working toward developing power-efficient machine learning-based hardware architectures for sleep apnea detection.

Radio frequency (RF) is ubiquitous in the surroundings from which energy can be harvested and utilized. Even for low-power sensors, RF energy harvesters can be utilized as primary power sources. However, power density of RF signals is very low and therefore building blocks of RF energy harvester needs to be designed carefully to maximize efficiency to gain suitable output power. The building blocks of RF energy harvester comprises of an antenna, an impedance network, a rectifier, a DC-DC converter, and a control circuit. Antenna receives electromagnetic signal and converts it into equivalent electrical signal. Impedance matching network matches impedance and ensures that maximum power is being transferred to rectifier. The DC-DC boost converter provides stable DC output voltage. Therefore, low-power circuit design technique is being implemented in this project to design individual blocks of RF energy harvester.

This project is targeting an effective and reliable integrated circuit solution for processing a sensor signal by using low power technique. A bulk-driven subthreshold opamp is proposed to be used in the signal processing unit to achieve the low power requirement and relatively simpler circuitry on-chip design. The next step is to integrate the opamp in the signal processing unit.