SENSOR BASED PROJECTS
VISUAL LIGHT COMMUNICATION
The exponential demand for wireless communication creates stress on the radio spectrum. It is becoming increasingly crowded and hence a need for an alternative medium of this is Visual light communication for communication in a closed environment. It consists of a transmitter which has light-emitting diodes(LED)s operating in the visible spectrum and a receiver that identifies the sent data through photodiodes. The major advantage of this system is that LEDs can blink at a very high rate faster than the human eye can detect. Hence, it doesn't cause any kind of disturbance.
WRIST WORN ACTIVITY TRACKER
This project aims to design a classifier that can classify a limited set of activities from a wrist-worn accelerometer data. The publicly available dataset is used to analyze its spectral information. By finding the proper filtering specifications, a Band-pass Finite Impulse Response filter with a Gaussian window is used to get the extract more meaningful data out of the noisy dataset. A Fine Gaussian SVM classifier is trained and designed using this filtered data, and the testing accuracy results are compared against the raw data.