Smart sensor system simulation using mpi infrastructure

Indoor Air Quality affects people lives in many ways such as efficiency loss, health problems and energy loss. Since energy costs has recently increased a lot, new technologies and materials are employed/used in buildings to isolate from cold/hot weather to cut the expenses. Opening the windows manually for ventilation at places where many people live together (school, hospital etc.) results in increasing at energy consumption. If enough fresh air is not taken into the buildings (classroom or patients? room), than the productivity has start decreasing. Especially, mental capacity of students will decrease dramatically. For this reason, measuring the air quality, besides temperature and humidity evaluating all together for using them in air conditioning control system will drop energy costs and increase manpower/brainpower efficiency. In addition to that, poor air quality causes health complaints such as headache, weakness, and eye itching. Measuring air quality and using the outcomes for controlling the air conditioning will affect human life, energy saving, and working efficiency positively. Indoor air quality can be measured continuously by using smart sensor systems and fresh air can be supplied to these places economically.A system can be evaluated by means of mathematical modeling, simulation or realization. Mathematical modeling contains analytical design details on paper and evaluation process take place on these details. Despite being the easiest one, error rate will be high comparing to the others. In simulation, a system that is not available and difficult to realize can be tested in many ways by writing computer software. This is why simulation is chosen frequently. In this study, a scenario based simulator system is implemented by software that imitates multiple sensors in each room of a building to measure air quality of the environment. Along with a simulator realization issues, data collection issues from multiple computers and data processing issues at a central place are discussed.The simulator is running on top of a Linux / MPI cluster that include 17 nodes and all the data between computers/processes are exchanged and collected in one computer through the MPI communication infrastructure. All data in the central node is collected by TCP/IP socket and a GUI helps managing the system. The GUI is written in C-Sharp by using Zed-Graph library to create on line charts. The simulator proves that realization of such a smart sensor system is feasible and the measurement results can be monitored easily.