Abstract: Location-aware is a key technology to reduce bandwidth waste which is a negative feature of P2P overlay network.The paper proposes a two-phase resolving method to improve the global convergence ability of network coordinates computation.This method first uses PSO on the initial solution set of a global optimization，then runs a local refinement in the vicinity of the global optimum to obtain the accurate network coordinates of each nodes.Through the two-step optimization，the depending on the remote landmarks is reduced and the applicable scope of the algorithm is increased.

Key words: P2P, network coordinates, Particle Swarm Optimization（PSO）

摘要： 位置感知的P2P覆盖网不但可以提高网络资源利用，而且可以减少带宽浪费。提出了P2P覆盖网中节点网络坐标的两段式最优求解方法。该方法首先采用粒子群算法对节点初始网络坐标值进行全局最优求解，而后在全局最优解的附近进行局部求精，得到节点的最终网络坐标值。通过两段式计算，提高了节点网络坐标值获得全局最优解的概率，降低了坐标计算对远距离参考节点的依赖程度，从而提高了算法适用性。

关键词: 对等网, 网络坐标, 粒子群算法