Bandwidth Analysis of the Ubiquitous Video Conferencing
Application
Neil Arellano, Aleksander Milshteyn, Eric Diaz, Sergio Mendoza, Helen Boussalis and Charles Liu
Structures, Propulsion, and Control Engineering University Research Center (SPACE URC),
California State University, 5151 State University Drive, Los Angeles, CA 90032, U.S.A.
Keywords: Ubiquitous Video Conferencing (UVC), Semantic Information System Network, Tuple Space, Client-server
Model, Bandwidth Metrics.
Abstract: The CSULA SPACE Center has endeavoured to develop the Semantic Information System (SIS) Network
for real-time project collaboration. However, the lack of uniform, real-time communication platform
application poses an inconvenience to the project collaborators, as they would be driven towards third-party
communication applications, such as Skype, MSN, Yahoo Messenger, etc. The use of these commercial
products does not incorporate moderation features between the network participants. In addition, these
applications have various conference capacity limitations and their simultaneous multi-device sign-in
feature can lead to possible concerns with information security (Alegre, 2009). The Ubiquitous Video
Conferencing (UVC) application has been designed specifically for the SIS Network in order to provide its
participants with dedicated multimedia channels and interactive communication. It is built on the integration
of Qt libraries, audio/video codec libraries of FFMPEG, and the image-processing library Open Computer
Vision. This paper presents the UVC application within the Semantic Information System Model and
focuses on issues related to real-time bandwidth regulation.
1 INTRODUCTION
The trends of increasing computing power,
affordability of hardware, and emergence of
embedded networked systems have enabled
businesses, researchers, and students the capability
of accessing the information through technologies
such as the Internet, technical software, and peer-to-
peer communication.
Although web-browsers, social network
applications, and voice/video communication enable
users to identify “birds of a feather” communities,
information is often scattered among various
networks often hindering the ability for information
updates to be synchronized in a seamless fashion.
Figure 1 shows an innovative SIS model of
software architecture. It provides facilitation and
management of client requests across a distributed
server network. It is particularity suitable for
information dissemination between project
collaborators, particularly for researchers, educators,
and students; specifically the use of the network by
groups where geographic location, time, and
computing resources limit information exchange and
collaborative efforts amongst each other.
Figure 1: Semantic information system model.
The proposed Semantic Information System
platform is best suited for educators, researchers,
and team-project members with common interest.
The platform allows these users to effectively
generate, analyse, and disseminate information. SIS
users will be able to “objectize” information nodes
of their projects, which will generate a hierarchical
tree structure to interrelate those nodes based on
their semantic meaning and relationship to each
other. These user-generated contents can be
301
Arellano N., Milshteyn A., Diaz E., Mendoza S., Boussalis H. and Liu C..
Bandwidth Analysis of the Ubiquitous Video Conferencing Application.
DOI: 10.5220/0004026103010304
In Proceedings of the International Conference on Signal Processing and Multimedia Applications and Wireless Information Networks and Systems
(WINSYS-2012), pages 301-304
ISBN: 978-989-8565-25-9
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
accessed, updated, and shared with other network
participants.
For the aforementioned SIS platform, a need for
communication interface was born. In order for the
project collaborators to interact with each other a
Ubiquitous Video Conferencing (UVC) has been
developed. It provides SIS users with real-time
interactive application that supports audio, video,
and textual means of communication for SIS project
collaborations.
The paper is organized as follows: The
Introduction of the SIS project establishes the need
for UVC system and is followed by the detailed
Facilitating Technology section. Then, the overview
of the Client-Server Architecture and UVC
Application Interface are discussed. The paper
concludes with the Performance Analysis section,
which presents bandwidth measurements.
2 FACILITATING TECHNOLOGY
The UVC application is utilized in conjunction with
several underlying technologies that constitute to the
development of the Semantic Information System.
Figure 2 shows UVC application within the multi-
server architecture, as well as structural component
interdependency of the SIS platform. A Tuple Space
programming paradigm allows initiation of client-
server requests. Tuple Space supports automatic load
balancing in parallel processing in a multi-
thread/multi-core server. Active Directory Service
System uses Tuple Space to provide server-to-server
communication infrastructure for distributed server
network configurations (Tolksdor, 2004).
Figure 2: Overview of SIS scheme and applications.
A PostgreSQL Relational Database Management
System performs data warehouse storage allocation
and unique key assignments to the content within the
network. The database keeps track of registered
users and their access privileges play an important
role in security.
Qt API is an open-source development
environment that is geared toward design of GUI
applications with extensive set of libraries available
to the programmer. To facilitate audio and video
compression, Open Computer Vision (OpenCV),
FFMPEG, and Speex libraries were integrated to
create the UVC Application. As mentioned before,
the UVC application provides its participants with
dedicated multimedia channels. Data streams
through those channels can be compressed or
encrypted based on different level of quality of
service (QoS) or security demands. The mechanism
of facilitating such needs is based on a
reconfigurable embedded subsystem. The related
technologies are beyond the scope of the paper and
will be published elsewhere.
3 CLIENT-SERVER
ARCHITECTURE
The UVC Application by default is designed to
operate in a client-server environment. The client-
server communication is handled through a Tuple
Space model. In a Tuple Space, an “OUT” command
deposits a tuple in, while an “IN” command retrieves
a tuple based on the keyword matching. The tuples,
described in XML, can be exchanged between a
client and its local server.
A Hyper-Threading architecture defines different
roles among the threads of each server. Figure 3
shows a controller and multiple worker threads. The
controller thread deposits and manages the tuple into
the Tuple Space. A worker thread retrieves a tuple
and performs the task based on the requests
described in a tuple. This delegation of roles allows
for efficient communication and task scheduling, as
well as automatic processor load balancing (Alegre,
2010).
Figure 3: Tuple space architecture with controller and
woker threads shown.
4 APPLICATION INTERFACE
The UVC is embedded in the Semantic Information
Network, where conference participants are able to
interact with each other. In project environments, a
WINSYS2012-InternationalConferenceonWirelessInformationNetworksandSystems
302
human moderation is necessary to organize and lead
the videoconferencing sessions. In moderated
sessions, an initiator of the videoconferencing is
automatically assigned with Forum Coordinator
privileges. When establishing a videoconference
link, the Forum Coordinator of the group will initiate
a session, thus relevant group members can join
through an authentication process. Members will
have their own user name and password, and will be
required to enter an additional verification code for a
specific video conferencing room. Its login panel
GUI is displayed on Figure 19. Upon joining,
members’ data such as the user name, IP address,
application capabilities (video/audio/text chat only)
will be logged. The Forum Coordinator will be in
charge of placing and lifting restrictions on each
participant such as who can use video/audio
channels. These operations will be done in the main
tab of the UVC software shown on Figure 4.
Figure 4: UVC Main GUI with video/audio, chat, finger-
painting pad, and voting system shown.
Individual participants will also be able to
regulate their own bandwidth, such as changing the
video resolution, closing certain incoming video
streams, adjusting their video frame rates
5 PERFORMANCE ANALYSIS
OF VIDEO/AUDIO CHANNELS
5.1 Video Performance Characteristics
The server bandwidth is determined by BW
s
= (P *
N) * S, where BW
s
is the bandwidth of the server, P
is the number of senders, N is the number of
receivers, and S is an average stream bit-rate of the
encoded Audio/Video content (Prasad, 2003). The
client bandwidth is calculated using the following
formula: BW
c
= P * S, where BW
c
is the bandwidth
of the client.
Network utilization is used to determine the
activity of the network port on the client system.
Three experiments have been performed in order to
find the average ratio of the network usage to the
maximum traffic. If the network utilization exceeds
the threshold of 50%, it causes system instability.
The sender is able to vary the FPS (frames per
second) rate and hence, influence the network
utilization. The first bandwidth experiment
determined the percentage of network utilization for
one video transmission with frame rates of 1, 5, 10,
15, 20, and 25 fps. Figure 5 displays a graph of
incremental bandwidth usage as different frame rates
are being established.
Figure 5: Frame rate versus network utilization.
This feature assists to the application users in
determining the appropriate frame rate settings
during various videoconferencing sessions.
Depending on the amount of conference participants,
as well as the number of active personal bandwidth
consuming applications, users can vary their video
frame rate, thus having an instant impact on their
bandwidth and video quality. As the frame rate gets
closer to the mark of 25 fps, the quality of the video
improves, while decreasing frame rate frees up
bandwidth resources.
The second bandwidth measurement experiment
was conducted in order to determine the average
amount of bandwidth utilized on a 100Mbps
network, with the transmission frame rate of 25 fps.
Figure 6 shows the network bandwidth usage with
25 fps rate for multiple UDP video connections.
Figure 6: Average Network Utilization versus Number of
Parallel Video Streams.
5.2 Audio Performance Characteristics
This section presents and discusses the experiments
performed to analyze the video and audio
bandwidth. In UVC, the video and audio channels
run independently from one another. The user gets to
decide which channels of communication to open or
close. The UVC GUI has various options that
provide users with an easy and efficient control
BandwidthAnalysisoftheUbiquitousVideoConferencingApplication
303
panel. Video frame rate control and selection of
audio transmission type are primary bandwidth
regulators that UVC GUI offers to the end user. A
client may choose from the three following types of
audio transmission: Pulse-Code Modulation (PCM)
raw audio streaming, Zlib codec, and Speex audio
codec. The experiments on audio performance were
done over IEEE 802.11g -54 Mbps network.
Audio transmission requires much less
bandwidth then video. However, in a collaborative
environment with multiple participants, who are
actively involved in performing various data
transferring operations, audio bandwidth
conservation can also be application-critical. To
offer audio compression solutions within UVC
application, the Zlib and Speex audio codec were
integrated to reduce the amount of bandwidth during
transmissions. Since Zlib codec has been already
incorporated within Qt library as its standard data
compression scheme, it was chosen as the first
compression option for audio transmission
(Roelofs). Zlib furnishes users with a lossless
scheme so the uncompressed output on the
receiver’s end is equal to the sender’s raw data
input. It was determined that, at least for this
particular sequence of samples, the data compressed
yielded a 29% of the original PCM data.
In order to have a more vigorous encoder, Speex
codec library was selected to provide optimal
compression ratio within the UVC application (Xiph
Org.) Upon successful implementation of the codec,
it was determined that on average Speex utilizes
only 0.08% of the network bandwidth. Speex
encoder on average is able to compress 75% of the
original PCM data. Table 1 shows bandwidth
utilization comparison between three methods of
UVC audio transmission, while Table 2 displays
bandwidth consumed by both audio and video
channels.
Table 1: Average Network Utilization for Audio
Transmission over IEEE 802.11g (54 Mbps).
Codec Type Ave. Network
Utilization (%)
Compression Ratio to
Raw PCM
PCM 0.31% 1
Zlib 0.22% 0.71
Speex 0.08% 0.25
Table 2: Average Network Utilization for Audio and
Video Transmission over IEEE 802.11g (54 Mbps).
Video at 25 fps with Audio
Codec Type
Ave. Network Utilization
(%)
Video + PCM 19.32%
Video+Zlib 19.24%
Video + Speex 19.08%
6 CONCLUSIONS
SIS participants are offered with a repertoire of
video and audio controls while using UVC
application. These transmission options provide
flexibility for bandwidth control, as the Forum
Coordinator and other participants can regulate their
network bandwidth in order to accommodate more
conference participants or enhance the quality of the
video/audio streams. The multithreading processes
of providing individual audio and video channels,
along with real-time video frame rate control and
audio transmission selection - add versatility to the
Ubiquitous Video Conferencing application. These
features enhance efficiency within the whole SIS
platform, as participants have direct control over
real-time communication channels. The performance
analysis of UVC application shows the benefits of
incorporated application options that allow flexible
real-time bandwidth regulation during
videoconferencing sessions.
ACKNOWLEDGEMENTS
Acknowledgement to NASA University Research
Center Program, GRANT # NNX08A44A.
REFERENCES
A. Alegre 2009. Aerospace Information Server. California
State University Competition, 2009. Structures,
Propulsions, and Control Engineering (SPACE)
Center – Research Topic: Ubiquitous Computing and
Embedded Architectures.
A. Alegre, S. Beltran, J. Estrada, A. Milshteyn, A. Lam, JP
Adigwu, C. Liu, H. Boussalis May 2010. Ubiquitous
Video Conferencing for Obectizing Contents in a
Semantic Network. 4th International Conference on
New Trends in Information Science and Service
Science (NISS2010), Gyeongju, Korea.
R. Tolksdorf, F. Liebsch, and D. Minh Nguyen, 2004.
XMLSpaces.NET: An Extensible Tuplespace as XML
Middleware.
R. S. Prasad, M. Murray, C. Dovrolis, K. Claffy, 2003.
Bandwidth estimation: metrics, measurement
techniques, and tools.
G. Roelofs, J, Gailly and M. Adler. ZLIB - A Massively
Spiffy Yet Delicately Unobtrusive Compression
Library.
Xiph Org. Speex - an Open Source/Free Software patent-
free audio compression format designed for speech.
WINSYS2012-InternationalConferenceonWirelessInformationNetworksandSystems
304
