Smart cities are constantly evolving ecosystems, offering individuals a safer, more intelligent, sustainable way of living. The success of smart cities relies on emerging technologies, such as Internet of things, artificial intelligence, high-definition video streaming, and mixed reality. Despite significant progress in these emerging technologies, the recent wireless networking infrastructure may be unable to support these technologies due to their large networking, computation, and storage requirements. In this talk, I will present frameworks and methodologies for empowering the wireless networking infrastructure to effectively support smart cities. In particular, we first focus on mobile edge computing systems and propose a deep reinforcement learning-based algorithm, which can effectively reduce the task processing latency. We also explore the potential benefits of joint resource sharing among mobile devices. Based on such resource sharing models, I will discuss how stochastic control, online convex optimization, and game theory can be employed for improving resource efficiency. Furthermore, specific applications (e.g., high-definition video streaming, virtual reality) are considered in our work to improve the service provisioning in smart cities under limited network resources. I will further provide visions of how machine learning techniques and wireless networking infrastructure can coordinate to make smart cities a reality.