欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为了提高P2P流量预测的精度，提出一种基于小波变换和回声状态网络的流量预测模型。将原始P2P流量分解为不同尺度的高低频分量，根据不同分量的流量特性匹配不同参数的ESN模型分别预测，将多路预测结果整合输出。通过对不同P2P软件的流量预测结果表明，该模型的流量预测精度可达到98％以上，显著优于传统的ESN模型和最小二乘支持向量机模型。
英文摘要：
To make improvement on prediction accuracy of peer to peer （P2P） traffic, a novel model based on the ensemble of wavelet transformation and echo state network （ESN） is proposed. The original P2P traffic is decomposed into multiple compo- nents of either low or high frequency, then each of these components is sent to a unique ESN model with suitable parameters which match their characteristics for prediction. The final prediction result of the original P2P traffic is considered as a weighted sum of the aforementioned multiple predictions. Experimental results show that the prediction accuracy achieved by the proposed model are always more than 98% which outperform the traditional ESN and the least squares support vector machine （LS-SVM） significantly.
同期刊论文项目
同项目期刊论文
期刊信息
位置：