ADAPTIT has completed its participation in the FP7 project ONTIC.
The project ONTIC, “Online Network Traffic Characterization” was successfully concluded in February 2017. The project developed innovative machine-learning (ML) algorithms for network traffic characterization and demonstrated their applicability in the context of dynamic network management http://ict-ontic.eu,
The consortium combined academic and industrial expertise, consisting of the Universidad Politecnica de Madrid, Politecnico di Torino, Centre Nationale de la Recherche scientifique, Toulouse, SATEC Spain, Ericsson Spain, DELL-EMC Spain and ADAPTIT Greece.
Three use cases were developed for exhibiting the power of ML algorithms to intelligently analyze raw network traffic data and distill useful information and actionable reports: on-line detection of anomalous flows based on unsupervised techniques, adaptive QoE management utilizing self-configured clustering and cluster quality algorithms and dynamic congestion avoidance featuring short-term prediction models based on neural networks.
The project also made available to the public a dataset with real network traffic traces (0.5 Peta Byte, header information only for privacy reasons) that has been captured from the network of a Spanish ISP during a period of 26 months.
ADAPTIT acted as a leader of the Big Data architectural activities, contributing to Spark technology aspects and the development of use cases while actively following the ML algorithms developed by the project for assessing their applicability from SME business perspectives. In particular, the work of ADAPTIT in the project included: tuning of the Spark distributed processing platform for optimizing application performance; provisioning of Spark clusters based on Google cloud services; solutions and scenarios for deploying and using ML-based analytics in service provider domains; techniques and Spark implementations for stochastic network traffic characterization and generation of synthetic traffic bearing the dynamics of the actual input traffic.
Overall, through its participation in ONTIC, ADAPTIT has strengthened its know-how and expertise in Big Data deployment, ML-based analytics development and visualization.