Information-Centric Networking (ICN) is gaining increasingly momentum, as an important direction of the future Internet architecture research. To study the impact of various cache policies on the overall performance of ICN network, we formulate the in-network caching problem of ICN into Mixed-Integer Linear Programming problem. Via analyzing the properties of the optimal cache assignment, we explore the impacts of the frequency and distance factors on cache policy performance. Furthermore, we design a novel cache policy named Least Benefit (LB), which takes into account the distance factor besides frequency. Through extensive simulations, we compare the performance of a variety of cache policies under various scenarios and configurations. It is found that LB outperforms a series of cache policies in terms of average hop number to get content.