● Analog Front End Design for non-intrusive long-term ECG Monitoring
In this project we designed a 0.5μm Analog Front End (AFE) for non-intrusive long-term ECG Monitoring.
In this project we designed a 0.5μm Analog Front End (AFE) for non-intrusive long-term ECG Monitoring.
In this project we showed the possibility of using carrier phase to detect soil moisture content from ground-reflected GPS signals. We used GPS-SDR-Sim to generate the raw RF data given coordinate or trajectory, time and sampling frequency. GNSS-SDR was used to process the raw RF data and MATLAB was used for analysis. We showed that it is possible to detect minute variations in the phase of the carrier to detect the soil moisture.
The project focuses on developing a compact and wearable system which detects the real-time stress level of a patient using motion compensated multi-channel multi-wavelength Photo-plethysmogram and Galvanic Skin Response, two non-invasive easily measurable physiological signals. An accompanied android app was also developed.
The simplified Microprocessor without Interlocked Pipeline Stages (MIPS) was designed using Cadence. Implemented the gate level structure followed by physical design that covered floorplanning, power mesh, clock tree synthesis, nano routing and post-routing timing optimization.
A cost effective 3-axis CNC milling machine for rapid PCB prototyping was developed from scratch. PCB Gerber files were converted to g-code using FlatCAM and the CNC was operated using Chillipeppr. This project was done in collaboration with PiLabs. (https://pilabsbd.com)
A CNN and LSTM model was trained using MS-COCO Dataset for generating image captions, which was later translated to braille using Arduino and solenoid switches. The model was trained on a virtual machine in Google Cloud Platform.
The project was basically a CNN based lip reading model. The MIRACL-VC1 dataset was used for training and testing. The pre-trained VGG-16 model was used for transfer learning.
A smart grid connected meter with Energy Consumption Scheduling (ECS) was developed with autonomous demand-side management using Game theory.
A prototype Zing Stump i.e. LED cricket stumps were developed using Reed switches and microcontrollers. RF communication was established between the base and the bails.
Single instrument automatic music transcription UI using FFT was developed in MATLAB.
A DC generator-motor combo to retrieve the brake energy and aid the rider while riding was developed. An SMPS and a charge controller for the battery was also designed.
Audio from a musical instrument was converted into a electrical signal whose frequency was measured using cascaded counter circuits with a magnitude adjuster.