Internet of Things (IoT) has been integrated into many fields of our lives, such as smart agriculture, supply chain management, vehicular network, medical health, and smart grid, etc. The development of the fifth generation (5G) network technology has promoted the upgrading of network infrastructure, making it highly interconnected, secure, reliable, and wireless, which enables modern IoT system to integrate computing, storage, and communication resources in a certain area to achieve the effect of improving productivity, efficiency, and security. Eventually, they achieve the purpose of improving people's quality of life. Thus, the future IoT system is bound to be a complex synthesis including blockchain, edge computing, artificial intelligence, and other technologies. In this talk, we not only analyze the future development trend, but also extract the optimization problems, such as computing distribution, delay minimization, power scheduling, and explain them in detail. Then, we discuss these optimization problems from two respects, based on traditional optimization algorithms (including algorithmic game theory) and based on modern machine learning models.