Recently, the concept of Internet of Things, referred to as IoT, has drawn a great deal of research attention for realizing an intelligent society. The IoT is expected to comprise millions of heterogeneous smart “things” having sensor terminals, which may collect various types of information. By sending these collected information via IoT, it is possible to construct many smart systems, e.g., automatic prevention of traffic jam and so on. However, the coverage of IoT and the capacity of its ground networks are, still, not capable enough to connect the numerous devices and terminals deployed all over the world. Therefore, there is an urgent need for effective data collection in such systems. In this paper, we focus on effective data collection by satellite-routed sensor system, which makes it possible to gather data from wide areas arbitrarily yet efficiently. However, multiple accesses from the numerous things to a satellite result in data collisions and increase the delay. For effectively resolving the problem of data collisions, we envision a new method, which utilizes a “divide and conquer” approach to collect data from the numerous things based upon demand. Also, we mathematically demonstrate how to optimize the operating time of our proposed. The effectiveness of our proposal is evaluated through numerical results.