Posts by Collection

portfolio

projects

Cloud/Edge computing support Infrastructure Development

Started from:

In this project, we developed a lightweight virtual cluster management system, named as Humilis, for enabling adaptable resource management in practice. For the easy transformation from cloud to edge with the network, Humilis decouples the control module, computing module and storage module in order to fit the uncertain requirements of edge computing. This design offers flexibility in allocating the virtual instances disk files in a shared disk manner or non-shared disk manner. In this architecture, a mobile user exploits virtual machine (VM) technology to instantiate customized service software on a nearby cloudlet rapidly, and then uses that service over a wireless connection.

Network Caching Framework

Started from:

In content distribution and 5G wireless networks, in-network caching has become an important solution for efficient data transmission in using network resources. Based on the acquisition of transport protocol features (e.g., TCP/IP, ICN, etc.), this project obtains cache metrics and proposes distributed content placement strategies for supporting network cache architecture.

Network Optimizations for 360-degree Video Streaming

Started from:

360-degree video stream has the problems of unnecessary data transmission and high quality QoE requirement in the AR/VR network field. The project aims to predict immersive video user behavior information and select appropriate video content based on network delay estimation as well as adjusting video bit rate.

publications

Towards Tile Based Distribution Simulation in Immersive Video Streaming

Published in IFIP NETWORKING 2019, 2019

Abstract

Abstract—In recent years, there has been an increasing at- tention on Virtual Reality applications, including 360 immersive videos, which typically consume much more bandwidth than traditional videos. However, only a small part of the produced data (denoted as Field of View or viewport) ends up being watched by the users due to the nature of 360 immersive videos. This negatively impacts the network, through a large increase in traffic and wasted resources. Hence, it is necessary to detect where the user is gazing, and what is the movement of the user’s head, in order to define a viewport-dependent streaming transmission strategy. As there are few datasets providing this information, it is crucial to define an easy-to-run model that generates such information. In this paper, we propose a tile- based simulation approach which can generate the distribution of the user’s behavior, and provides information that can be used to optimize future view-dependent streaming protocols. We first characterize the users’ viewport pattern from datasets gathered from real users by decomposing the 360 stream into tiles and analyzing for each tile its frequency and its time interval distribution. Then, we devise a simulation model by combining a Markov Model for tile transition, and beta distribution for time interval prediction. The results show that the simulation tool is able to characterize the tile sequences of users, which performs closely when compared with the empirical results.

Recommended citation: Dongbiao He, Jinlei Jiang, Guangwen Yang, Cedric Westphal and J.J. Garcia-Luna-Aceves. Towards Tile Based Distribution Simulation in Immersive Video Streaming. IFIP NETWORKING 2019

talks

Talk 1 on Relevant Topic in Your Field

Started from:

This is a description of your talk, which is a markdown files that can be all markdown-ified like any other post. Yay markdown!