"Over the last few years, the proliferation of Internet services has lead to a dramatic increase of network applications. However, reliable mechanisms for effective traffic profiling of network applications, and classification of the traffic flows according to the applications that generate them remain elusive. This lack of mechanisms combined with the fact that most modern applications intentionally or not “hide” their generated traffic has a number of adverse consequences: (i) it leads to inefficient management of network resources and infrastructure, (ii) it exacerbates the problem of security, and ultimately, and (iii) it induces poor end-user experience of the provided services, discouraging adoption of advanced technologies and Internet services. Our goal in this project is to advance the state of the art in the area of application identification by investigating algorithms and implementing a system that allows robust and automated application classification. The proposed methodology will be based on capturing the communication patterns of network applications in graph structures without inspecting the user data, respecting in this way user privacy concerns. This is attractive, since these properties are hard to obfuscate without modifying significantly the application itself, while at the same time the methodology succeeds even in the case of encrypted data. We believe that the collaboration proposed in this project has all the potential to bring it to a successful completion. Combining (i) the applicant’s knowledge of, and contributions to the state-of-the-art in, the area, with (ii) the extensive experience of the host in network measurements, will allow us to explore novel algorithms and solutions for the problem of traffic classification. Beyond its research merit, the proposal will significantly add to the applicant’s long-term career development and academic experience by setting this work as the point of reference in the area."