The proliferation of the Internet of Things (IoT) has paved the way for the future of smart cities. The large volume of data over the IoT can enable decision-making for various applications such as smart transportation, smart parking, and smart lighting. The key to the success of smart cities is data collection and aggregation over the IoT. Recently, crowdsensing has become a new data collection paradigm over the IoT, which can realize large-scale and fine-grained data collection with low cost for various applications. For example, we can leverage the power of the crowd to build a real-time noise map with microphones on smartphones. Despite the advantages of crowdsensing and the IoT, there are many challenges to utilize crowdsensing over the IoT for smart cities, such as how to allocate tasks to appropriate users to provide high-quality sensing data, how to incentivize users to participate in crowdsourcing, how to detect the reliability of the crowdsourced data, and how to protect the privacy of users.