There is no doubt that smart city applications dramatically increase and the need for smart cities in our modern life becomes a demand. Smart cities will enable various applications and introduce many market innovations. However, the dramatic increase in wireless devices and network traffic puts many constraints in designing such networks. To this end, we provide to develop a reliable smart city system that enables various heterogeneous applications and provides the communication infrastructure for the expected enormous number of wireless devices. The proposed system is scalable so that the increase in network traffic will be supported with no degradation in network performances. The system deploys edge computing servers and artificial intelligence. In this study, we simulated a model for the structure of a smart city based on heterogeneous edge computing and defined evaluation parameters. Finally, according to the analysis of the results will be developed a prototype for the practical implementation of the selected method using a specific example is based on a neural network algorithm to generate forecasts of Internet of Things traffic activity.