Edge-based Collaborative Sensing Toward Learning Human Activities in Real-time
AdaSift
Fine-grained Indoor Air Quality monitoring: Development, Calibration, and Prediction under the project IMPRINT , Under the Guidance of Prof. Subrata Nandi and Dr. Sujoy Saha,Department of CSE , NIT Durgapur [Jun 2018 - Oct 2020]
"Pollution is the introduction of harmful materials into the environment”
The rapid increase in industries and transportation help us to develop our society and lead us to a top-class living, but at the same time, it is pulling us towards the world of adulterated environment. It has been monitored that the environment of indoor is nine times more polluted than that of outdoors. So, we design and develop a system to monitor the air quality of indoor as well as outdoor. We also do the estimation of indoor air pollution without dense placement of air quality monitoring stations(AQMS).
To validate the EMD data, we calibrate the EMD because some times there exist sensor drift. We optimize the sensor errors using soft calibration. Though it reduces error from the dataset but needs huge energy.
So here we propose an energy-aware calibration technique to calibrate the EMD. After the calibration, we placed the EMDs sparsely. We estimate the pollutant concentration where no EMD is present. An accuracy of 95.86% has been obtained by using a multilayer perceptron.
At the time of estimation, we also face the challenge that where we place the EMD to monitor the air quality of a building. It is not feasible for us to deploy EMDs in each room of each floor of each building because the number of the floor of a building is increasing due to the land depreciation, the increase in population, coal, and fuel mines. So optimal placement technique is needed. Placing the desired number of devices in optimal rooms is a nontrivial problem. In this work, an optimal placement strategy of air quality monitoring devices has been given. This work infers the optimal locations of devices to be placed with the desired number of monitoring devices.
Road Surface Event detection using heterogeneous sensors under the project of CityProbe ,NIT Durgapur [Jun 2018 - Dec 2018]
A low cost arduino based portable and wireless device which is capable of detecting the road damping, speed breaker, pot holes and so on with the help of low cost sensors. The mentioned device also can inform the driver about the road condition. Not only this, but also this device is capable of informing about the vehicles around it.
Human Activity Reorganization using ultrasonic Sensor, PIR Sensor under the project DISARM, Department of CSE and CA, NIT Durgapur Funded by Media Lab Asia, Govt. Of India [ Dec 2015-June 2016].
To innovate smart homes with time saving and smart services including energy saving methods, automatic operation of electric appliances etc. Automatic Human activity recognition also plays a key role in it. There are several daily human activities that occur and there are various methods to trace them. But, mostly the issue lies in costly deployment of sensors or hampering the privacy of an individual by using cameras. Hence, in my work, with the help of various clustering algorithms we could observe and differentiate certain human activities very clearly using ultrasonic sensors.
We obtain the data from the sensors and divide them in uniform time slots and then, refine them using various clustering algorithms so as to obtain fruitful results. We here observed for whether a single person is sitting or standing or a group of people are sitting and having a discussion or they all are standing in a group. This is a cost effective proposed method and hence, provides 85% and above accuracy.
Classification of Aromatic rice samples using FT-NIR spectrometer and MATLAB (s/w) from C-DAC Kolkata. [Jun 2015 - Jul 2015]