BitTorrent is one of the most well known peer-to-peer file sharing applications, accounting for a significant proportion of Internet traffic. Current BitTorrent system builds its overlay network by randomly selecting peers, a fact that has the potential to seriously handicap both individual performance and generate a significant amount of cross-ISP traffic. In this paper, we propose a novel peer selection algorithm called UTAPS which could selects peers within small hop counts and low round trip time (RTT) range by utilizing the knowledge of underlying topology. Consequently, the proximities among peers in the P2P overlay network are enhanced. Simulation results show that the UTAPS algorithm can achieve better individual performance in sense of low download time and reduce the traffics which are injected into ISP backbones.