A Dew Computing Architecture for Smart Parking System with Cloud Image Recognition Service

In the smart parking system, knowing the real-time status of parking spaces can resolve many service problems. In the current trend, a well-deployed cloud image recognition system can fulfill this requirement in which the recognition model's training and adjustment have been delegated to the cloud service provider. Consequently, the scalability and economy of the system have been improved, and the business owner also releases their burden on high-tech system management. However, cloud service has a constraint on Internet connection dependence. Hence, we propose an innovative cloud-dew architecture for the smart parking system. The advantages of this architecture are: improve the micro-service property of dew computing, function independence without Internet, automatically service proxy to increase the horizontal and vertical scalability, and dedicated to smart parking system development.