This disclosure describes techniques for detection of duplicate data transfers over a backbone network based on time series correlation of the data traffic. Per techniques of this disclosure, a time series of data traffic is generated. Correlation of the time series of traffic flows from different servers is performed to determine duplicate data transfers. A correlation coefficient between different segments of traffic flow time series is determined. Segments of data traffic with a correlation coefficient that meet a threshold correlation coefficient (correlation coefficient close to 1) are identified as likely duplicate transfers. The described techniques can be utilized to detect duplicate data transfers within a backbone network. The techniques enable removal of duplicate data transfers leading to substantial cost savings and can help alleviate network congestion.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Anonymous, "Inferring Duplicate Data Traffic in Backbone Networks", Technical Disclosure Commons, (December 16, 2020)