PROJECT TITLE :

Accurate Recovery of Internet Traffic Data Under Variable Rate Measurements - 2018

ABSTRACT:

The inference of the network traffic matrix from partial measurement information becomes increasingly vital for various network engineering tasks, like capability coming up with, load balancing, path setup, network provisioning, anomaly detection, and failure recovery. The recent study shows it's promising to more accurately interpolate the missing knowledge with a three-D tensor as compared with the interpolation ways primarily based on a two-D matrix. Despite the potential, it is troublesome to create a tensor with measurements taken at varying rate in a practical network. To address the problems, we have a tendency to propose a Reshape-Align scheme to form the regular tensor with data from variable rate measurements, and introduce user-domain and temporal-domain issue matrices that take full advantage of features from both domains to translate the matrix completion drawback to the tensor completion downside primarily based on CANDECOMP/PARAFAC decomposition for additional accurate missing knowledge recovery. Our performance results demonstrate that our Reshape-Align theme will achieve significantly better performance in terms of many metrics: error ratio, mean absolute error, and root mean square error.