为提高端到端可用带宽的预测精度,提出了一种基于自治域(AS)拓扑的可用带宽预测算法T-aware.该算法结合网络的AS拓扑信息,为符合超度量空间的节点和不符合超度量空间的节点分别选择测量点,对2类节点分别进行预测.在HP scalable sensing service数据集上的实验结果表明,T-aware的预测精度与PathGuru算法相比可提高10%以上. To improve the prediction accuracy of end-to-end available bandwidth, an adaptive bandwidth prediction algorithm T-aware based on Autonomous Domain (AS) topology is proposed, which combines the AS topology information of the network, The nodes in hyper-metric space select the measurement points separately and predict the type 2 nodes respectively.The experimental results on the HP scalable sensing service dataset show that the prediction accuracy of T-aware can be improved by more than 10% compared with the PathGuru algorithm.