social media traffic classification in encrypted traffic.

2018 Research Internship Project

---

Faculty Name

Farah Mohammadi

Project Title

social media traffic classification in encrypted traffic.

Project Description

Network traffic packets carry data belonging to a variety of applications. Classification of traffic helps network operators to identify specific applications and protocols that exist in a network, which can be useful for many different purposes such as network planning, application prioritization for QoS guarantees, and policy deployment for security control. In addition, traffic classification is used for trend analysis, application performance analysis, fault analysis, and network/security forensics. The increasing deployment of encryption in network protocols and applications poses a challenge for traditional traffic classification approaches. Approaches such as deep learning algorithms cannot accurately classify network traffic when packet payloads are encrypted because their pattern matching algorithms do not have access to the raw packet payload. As a result, we seek to leverage new approaches which can more accurately classify and identify applications in encrypted traffic. In particular, Machine Learning (ML) approaches have shown promise in this area. While researchers have studies and shown the viability of using ML approaches to identify applications such as Skype, accurate identification of social media applications within encrypted traffic remains a challenge. Consequently, there is an urgent need to develop machine learning algorithms for classification/detection of specific application types in encrypted traffic. More specifically, in detecting social media class of traffic along with the sub-class when traffic is encrypted and there is no access to the packet/traffic payload. A key challenge is to ensure that any proposed approach can classify social media applications with high accuracy in near real-time is a challenge. The project will explore the viability and applicability of ML algorithms for social media traffic classification in encrypted traffics.

Student Responsibility

1. Identifying the traffic features that are representative for a particular application. A detailed analysis of each application will be undertaken. 2. Identifying a suitable machine learning algorithm that yields accurate result for this target. 3. Creating a data-set by running social media applications in a laboratory environment.

Specific Requirements

1. A good background in C/C++/Matlab. 2. Be Professional in Algorithm designing. 3. Computer Networks. 4. Machine Learning.