The bandwidth of the network is precious. As cloud computing provides powerful computation power and more and more devices are connected to the Internet there is a need to seek for methods to reduce the traffic. Fog computing was proposed to adding the computation and storage ability to the edge devices. The requests from the end device can be pre-processed or even responded at the edge device. In this work, the ability of Web Resource Caching is added to the edge device to serve as a caching proxy server. In order to obtain more caching storage, the end devices are also exploits to provide some caching space. As the whole caching storage is enlarged, large resource like media files are able to be cached near to the end devices once any of the end devices in the same network just recently requested it. The result shows that the proposed architecture has better downloading latency compares to the single caching proxy approach.