Classifying Encrypted WiFi Traffic Using Deep Learning Methods
- Publication Type:
- Thesis
- Issue Date:
- 2022
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
In this thesis, the goal is to classify encrypted WiFi traffic using deep learning methods. Firstly, we investigate the possibility of making useful inferences from passively observed WiFi traffic that is encrypted at both the transport layer as well as the MAC layer. In the first work, we demonstrate the possibility of identifying video content using different deep learning models. Secondly, not limited to video streaming, other types of traffics(e.g. web and audio streaming) need to do classification due to the purpose of service management. In this work, we propose a novel hierarchical classifier that can make coarse-grained predictions (e.g. web, video, or audio) as well as fine granular predictions (e.g. content providers/platforms and exact content). Finally, we investigate how to generate WiFi traffic samples by category automatically. In this work, we propose two novel generative models, namely infinite Gaussian mixture auto-encoder and the infinite mixture of infinite Gaussian mixture auto-encoder. Infinite Gaussian mixture auto-encoder is a variant of variational auto-encoder with an infinite Gaussian Mixture model as the prior distribution of the latent variables. infinite mixture of infinite Gaussian mixture auto-encoder is a variant of variational auto-encoder with the infinite mixture of infinite Gaussian Mixture model as the prior distribution of the latent variables.
Please use this identifier to cite or link to this item: