Environmental Monitoring using Sensor Based Wireless Embedded Systems and ANN

A comfortable environment can increase the productivity in multi-folds. So it is important that the environment variables, such as temperature, humidity, light intensity and air quality (gas/smoke) are continuously monitored and adjusted to maintain a comfortable working environment with suitable threshold values on each of the variables depending on the conditions and quality of services required. Pervasive computing is one of the technological solutions that provide services in such an environment. In order to handle the challenges in monitoring, adaptability and maintenance of the ambience for a comfortable environment a sensor based wireless embedded system is designed using machine learning algorithms. In this framework the machine learning algorithms are embedded at the local level for decision making to filter the noisy and unwanted data during continuous monitoring and adaptability. The proposed model is implemented using an Intel Galileo Gen 2 board, sensors configured with the board are temperature, humidity, light, sound and gas sensors. Machine Learning algorithms are designed using Back Propagation Neural Networks which are deployed along with embedded software. Sensor data collected from the environment are used as training dataset for the Machine Learning algorithms with suitable decisive parameters. Back Propagation Neural Network is implemented to perform tasks such as predictions of the environmental parameters, expected threshold levels and averages. The main advantage of Back Propagation Neural Network is that it can fairly approximate a large class of functions. In formulating the ANN-based predictive model; three-layer network has been constructed. The Neural Network is trained and tested and the accuracy of the algorithm is determined. The neural network based prediction is integrated in real time monitoring, analysis, and control system for environmental conditions. Thus this project provides a prototype of a smart environmental monitoring system which can analyze large amount of environmental data and predict in real time to support decision-making and related tasks.