Machine Learning based framework to predict the Network Usage in Smart Parking Applications

The automotive industry is growing daily, as many people prefer personal vehicles for many reasons. Due to the increased usage of private cars, especially in growing cities, parking space availability has become short and poses a big challenge. It also leads to further cascaded issues like traffic congestion, wasting time in finding free space, polluting the environment, and especially unnecessary wastage of fuel. Therefore inventing a smart parking system has become the need of the hour. Many researchers have attempted to design solutions to this issue by utilizing state-of-the-art technology in prominent areas like Wireless Sensor Networks, Cloud Computing, and Internet of Things. However, there is still scope for improvement in smart parking system performance. In this research work, we are simulating a smart parking system to get the parameters like network usage, transmission time, number of areas, and number of cameras used. The simulator works better for a small number of areas, but for larger-scale simulations, it takes more time. Hence, we propose a framework that can predict/analyze the performance of smart parking systems at an enormous scale using an ML algorithm. The experimental results show that predicting network usage of large-scale smart parking systems using an ML framework is 1500x faster than the simulation time of the CloudSim simulator.