Machine Learning Empowered IoT for Intelligent Vehicle Location in Smart Cities

Abstract

Intelligent Transportation System (ITS) can boost the development of smart cities, and artificial intelligence and edge computing are key technologies that support the implementation of ITS. Vehicle localization is critical for ITS since the safety driving and location-aware serves highly depend on the accurate location information. In this article, we construct a vehicle localization system architecture composed of multiple Internet of Things (IoT) with arbitrary array configuration and a large amount of vehicles in smart cities. In order to deal with the coexisting of circular and non-circular signals transmitted by vehicles, we proposed several vehicle number estimation methods for non-circular signals. Based on the machine learning technique, we extend the vehicle number estimation method into mixed signals in more complex scenario of smart cities. Then the DOA estimation method for non-circular signals based on IoT is proposed, and then the performance of this method is analyzed as well. Simulation outcomes verify the excellent performance of the proposed vehicle number estimation methods and the DOA estimation method in smart cities, and the vehicle positions can be achieved with high estimation accuracy.