The idea of connecting things to the Internet (IoT) has become the hottest topic today. By 2020, twenty-five billion things are expected to be connected to the Internet. However, the host to host IP protocol, cannot be compatible with the situations of this network of things. Situations such as resource constrained devices (low power processor, limited in data storage, limited in electrical energy), intermittent reconnection, and a very high number of things. Named Data Networking (NDN) is an appropriate candidate for the future Internet of things because NDN shifts the network model from host-centric to information-centric, and with the unlimited namespace, data can be copied anywhere in the network. This in-network caching helps information spreading and also intermittent connectivity in IoT. NDN works on the base of the requested content's name instead of the physical location of the content. Content requests are routed towards the node that is able to respond to them. Therefore, content requests are routed by a name-based routing table. Caching affects the performance of NDN because it enables NDN to deliver contents on-time and efficient. When the cache becomes full, replacement policies are used to replace a currently stored content with the new one. Information of the FIB table is used in this paper to propose a new cache replacement policy. In the proposed method, the content that can be retrieved fast is removed from the cache to provide the space for the new content's storage. Simulation results indicate reduced cache redundancy, reduced average round-trip time, and also the same overall cache hit rate for the proposed method compared to the LRU policy.