International Journal of Engineering and Technical Research (IJETR)

ISSN: 2321-0869, Volume-2, Issue-2, February 2014

Modified HLS Model for IPTV Streaming

S. Japertas, T. Baksys, T. Zelvys

sampling (segmentation) principle suggested by Apple that

Abstract Present basic IPTV streaming technologies (as allows audiovisual data stream provision to the customers.
plain Multicast) become too slow, static and dont satisfy users This technology allows segmentation of multiple data flow
requirements. To make streaming more mobile, there was tested and it can be provided to the users in Open Source format.
Apple Inc. (Apple) HLS model. Regarding to the test results
there was created a modified model, which has better
characteristics of traffic shaping, bandwidth load balancing and II. STANDARD APPLE MODEL
mobility. Collecting user statistics and understanding their The concept implemented by Apple is reflected in the block
habits allowed creating an adaptive streaming way that could diagram Fig. 1. The main task of Apple HLS model is to
transfer data to users collectively without wasting the traffic.
discretize stream in real time and to transmit received
Every user is connected to a unicast-based cache server, which
collects user requests and connects to the main HTTP server. information to the user by using HTTP(s) protocol. The flow
Cache servers start downloading the information form HTTP of continuous stream in POSIX operation system is shown in
server only on demand and only the information that is 1 4 blocks. Multimedia streams are formed of audio-visual
requested. The downloaded information is cached in the Cache information that is transferred to the HLS server where it is
servers memory and can be shared among the parallel clients, buffered. In the software segmentator the stream is cut into to
limiting the traffic only to the one channel is broadcasted.
indexed files, which are transmitted to the servers dynamic
random access memory (RAM). RAM is used for short-term
Index TermsDigital multimedia broadcasting, IPTV,
Unicast, Cache memory data storage when data recording and reading is performed at
a high speed.
It is very important to ensure a high speed of data recording
I. INTRODUCTION and reading when disk arrays of the server become physically
Problems of telecommunication services development are incapable of manipulating the obtained data while processing
experienced recently. The problems are caused by outdated of high-quality HD (High Definition) video material. Later
communication technologies that require a large investment the information is transferred to the hard disc array of HTTP
to be replaced, a homogeneous network structure and specific server.
software solutions. The server performs streaming procedures to the clients in
These trends especially occur in networks that are used for HTTP protocol by establishing TCP [2][3] sessions.
audiovisual data streaming, which traditionally uses the This Apple concept realization is only compatible with
Multicast signal transmission technology. This information Apples high-end devices such as iPhone, iPad, iPod Touch,
transfer technology greatly restricts data accessibility to the MacBook, which means that the customer is obliged to use the
consumer by setting very high standards of service quality hardware and software created only by this corporation.
requirements, which obliges the audiovisual service provider
to ensure very high network permeability. Along with
increasing consumer needs, service providers are
experiencing service accessibility, supply and large data flow
problems, which are tried to being tackled by different
companies.
The review of transfer technologies that were made by the
IETF team (Internet Engineering Task Force) revealed that
the HTTP Live Streaming (HLS) concept of audiovisual data
transmission over IP networks. HLS describes the principles
of continuous IPTV stream sampling and transfer by HTTP
protocol implemented by company Apple is the most
universal [1].
This work presents the solution which uses HLS stream
Fig. 1. Block diagram of the Apple HLS model: 14 blocks is
POSIX system transmission, 56 blocks is networks system
Manuscript received February 20, 2014.
Saulius Japertas, Telecommunications Department, Kaunas University transmission
of Technology, Kaunas, Lithuania, Mobile phone No +37068525830.
Tatutvydas Baksys, Telecommunications Department, Kaunas This model limits the development of other devices and
University of Technology, Kaunas, Lithuania, Mobile phone No
+37061447065.
prevents Internet market growth.
Tautvydas Zelvys, Department of Multimedia Engineering, Kaunas
University of Technology, Kaunas, Lithuania, Mobile phone +37037300530

147 www.erpublication.org
 Modified HLS Model for IPTV Streaming

III. MODIFIED HLS MODEL network modules. IP communication channels have been
The HTTP Live Streaming traffic management software designed in several stages.
was ported into open source format, using C programming The first stage included the activation of set protocol to
language, compiled with the GCC compiler. Most of the establish a connection between the system modules. The NFS
multimedia signal streams between the channels were (Network File System) protocol, which uses ISCSI (Internet
replaced by protocol communication channels (SSL Small Computer System Interface) extensions, was chosen for
encryption was used). It has also been adapted to the stream data exchange between systemic POSIX subsystem and
distribution mechanisms that allow decreasing the HTTP server. The aim was to facilitate and expand the
transmission traffic and incorporating audiovisual (video part possibility of the connection between HTTP server and the
of the stream is dominating [4]) data stream into general IP subsystem of segmentator.
network flow. This enabled the IP stream to be compatible This was done by means of programming, creating NFS
with equipment of various manufacturers and get much better access to the RAM subsystem (after giving it networking ports
systemic and network load characteristics. Modified HLS and activating the software to exchange data in NFS protocol)
model is shown schematically in Fig. 2. and programmatically connecting to the HTTP port of the
Intermediate servers are installed in the modified model - protocol corresponding server to perform the task of
Intermediate (Cache) servers, which distributes traffic to each broadcasting. This connection feature can only work if the
other. These servers work in peer-to-peer mode, thus reducing web-enabled devices (communication lines), which transmit
the routes of data accessibility to the users and reducing the information, are of extremely high-quality, high-speed
response times. (channel data rate of around 1 Gbps) and of a high bandwidth.
This enables the user with possibility to connect any of the To ensure the stability of Information transmission the
closest intermediate server which updates its database with restriction for the core operations of software channels
necessary parameters in real-time. So, in this case, the data occurred. This restriction is introduced to the protocols to
flow to the intermediate server and between intermediate prevent the protocol errors in packet networks. Due to the fact
servers is not transmitted constantly but according to the adaptive interconnects (between blocks 5 6 and 6 7) were
users request. Data exchange is performed between network used, which allows the Cache servers to make data transfer
devices that are the closest and have the best bandwidth and only on client request. To solve the problem of adaptive
accessibility parameters. They are used to send information to connection the cache (temporary storage roaming) method
an intermediate server which can be addressed by the user to was used. This solution was implemented with reprogrammed
receive required information. Squid software package that was installed in cache
Relations in this modular system are multi-type POSIX intermediate server and intended to temporarily store Web
[5] subsystem signals (between blocks 1 4) and IP packets page metadata (images, tables, etc.).
data (between blocks 4 7). POSIX subsystem signals are In this case the cache intermediate server operates not as a
formed in the POSIX kernel bottom layer, they are local and standard server, but as rebroadcasting server adapted to
not transmitted to the remote servers. Meanwhile IP packet audiovisual broadcasting and able to balance network load
data is transmitted via telecommunications network by accumulated between adjacent servers. The client while
Ethernet signal. So HLS module becomes broken up into two connecting to an intermediate point uses its granted resources
parts that use two types of signals this way optimizing the use without directly increasing the load HTTP server. If quantity
of module resources and ensuring the stability of N of concurrent users are watching the same channel, an
performance. intermediate server calls HTTP server downloads the
necessary audio-visual information only once and distributes
it to N users. This network management principle allows the
Unicast signal transmission to compete with the Multicast
method of group streaming.
The second stage was to improve the traffic safety
parameters. For this purpose the HTTP server and the
intermediate servers were linked to each other through SSL
tunnels [6]. After the encryption of data exchange session via
SSL tunnels, the possibility to take unauthorized audiovisual
content information was significantly reduced.

IV. RESULTS
Stream and user parameters. The purpose of this
Fig.2. Scheme of the Modified HLS model: 14 blocks is research was to measure and compare several TV data
POSIX system transmission, 57 blocks is networks system streams parameters in various unmodified and modified HLS
transmission model locations. Primary research data is presented in Table
1.
Developing software the usability of Internet channels Multicast streams. Multicast streams survey was
connections was taken into consideration in order to create conducted using HLS models which were discussed above by
autonomously functioning system which consists of remote connecting them to the GPON optical network. Multicast

148 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-2, Issue-2, February 2014
stream was measured at the HLS model points that are temporarily indexed Cache server data.
presented in the Fig. 3.

TABLE I. USER AND STREAM PARAMETERS

The number of channels in Mcast signal 62
The number of watched channels during 10
the study
Number of users 50
Channel stream, Mbps 2
Duration of the study, h. 168

The specifics of Multicast signal transmission is in the fact Fig.3. Schematic of streams measurement points layout: 1
that the required information is sent to all clients at the same GPON network node, 2 measurement point Nr. 1, 3
time. However, this broadcast method can be applied only to IPTV data stream, 4 network distributor, 5.1 5.n
fully homogeneous network, which has no additional network customer stream ports, 6 measurement points No. 2.
nodes. The network topology of GPON, FTTH, etc.
telecommunications is usually of ring architecture, where The Cache server activity is represented by distribution of
network information is distributed to clients through network the amount of information transmitted in real-time (curve 5) is
distributors and such networks don't have adaptive stream presented in the Fig. 5. By comparing unmodified and
control subsystem. The continuous stream is transferred by modified HLS streams, it can be seen that the stream of is
this network is sensitive to data signal delays, different traffic smaller in the case of modified HLS data transmission model.
bandwidth changes, that is why all the available TV channels By selecting the optimal Cache server settings even better
information is sent continuously at the time. Such a network results created by modified HLS can be expected.
data stream measurements were carried out at the points of
measurement No.1 and No.2, and the results are presented in
Fig. 4. Curve 1 and 2 reflect the essential Multicast signal
transmission characteristic the stream is transmitted to the
transfer server and then to the client at almost the same rate
with 3Mbps within the range.
Unicast streams in unmodified and modified HLS
models. Two types of measurements using Unicast streaming
technology were performed. First type of the case is
unmodified HLS measurements were done in the data bus
between blocks 5 and 6 (see Fig. 1), second case is modified
HLS measurements were done in the data bus between
blocks 5 7 and blocks 6 7 (see Fig. 2). The sum of recent
measurement is identical to the conditions of measuring point
Nr.2 in the Multicast broadcasting case. After making the
measurements Unicast traffic speed and transferred data
volume dependency were obtained, at the unmodified and
Fig. 4. The distribution of network random realization data
modified HLS broadcast methods. Fig. 4 and 5 show the
traffic in real time
curves of stream distribution at the observation time, which
characterize the features of these broadcast [7].
In the case of unmodified HLS model the stream was
transmitted by Unicast transmission method. The streams
distributed highly dependently on the amount of customers
connected to the server. Since during the study there were 50
clients with traffic of 2 Mbps for each user, the stream formed
by all users was of 100 Mbps (it is presented in the Curve 3).
Curves 4 and 5, shown in Fig. 4, present uneven streams
distribution in the time interval in the case of modified HLS
model. This unevenness is obtained due unequal TV channels
viewing in a week period. During the study, it was observed
that clients usually are watching 10 TV channels (forming
most popular 10 channels). This client behavior allows
optimizing the communication channels and network devices
work according to the amount of information needed. In the
modified HLS model this optimization is performed by Cache Fig. 5. The distribution of network random realization data
server, which disables clients from creating new sessions with amount in real time.
HTTP server and allows using the already existing

149 www.erpublication.org
 Modified HLS Model for IPTV Streaming

V. CONCLUSIONS University of Technology. The area of interest is C/C++ programming,

embedded systems, parallel computing solutions and evolvement.
1. The user can receive data using modified HLS model by
Unicast transmission method.
2. This modified HLS streaming service accessibility and
equipment supporting overcomes Multicast transmission
method.
3. Operating HLS model allows the use adaptive streaming
cache mechanisms, which according to statistical criteria.
4. HLS allows to maximally optimize the stream
distribution to the users and allows the customers who use
mobile or small speeded connection to access the data.
5. Modified HLS model allowed to increase the security of
data sent.

REFERENCES
[1] R. Pantos, Ed. W. May (October 14, 2013). HTTP Live Streaming
[Online]. Available:
http://tools.ietf.org/html/draft-pantos-http-live-streaming-12.
[2] J. Summers, T. Brecht, D. Eager, B. Wong (2012). Methodologies for
generating HTTP streaming video workloads to evaluate web server
performance [Online]. Available:
https://cs.uwaterloo.ca/~bernard/nossd47.pdf.
[3] I. Marsic. (June 11, 2013). Computer Networks: Performance and
Quality of Service. Rutgers: The State University of New Jersy
[Online]. Available:
http://www.ece.rutgers.edu/~marsic/books/CN/book-CN_marsic.pdf.
[4] U. Jennehug,T. Zhang, Increasing Bandwidth Utilization In Next
Generation IPTV Networks, International Conference on Image
Processing (ICIP) 2004, vol. 3, pp. 2075-2078.
[5] POSIX. Basics for Users and System Administrators, Fujitsu
Technology Solution GmbH, 2009.
[6] Ross J Anderson, Security Engineering: A Guide to Building
Dependable Distributed Systems, Indianapolis: Wiley, 2008.
[7] X. Hei, C. Liang, J. Liang, Y. Liu, Keith W. Ross, A Measurement
Study of a Large-Scale P2P IPTV System, IEEE Transaction on
Multimedia, vol. 9, no. 8, December 2007, pp. 1672-1687.

Saulius Japertas, Assoc. prof., head of

Telecommunications department of Kaunas University of Technology. PhD
was received in 1991. The area of interest is Wireless Telecommunications
and Telecommunications (included Cyber) Security. Have published papers
in various fields of telecommunications and information technology.

TautvydasBaksys, M.Sc. of
Telecommunications, Virtual Center of Excellence Telecommunications
department of Kaunas University of Technology. The area of interest is
POSIX systems, Cloud and Virtualization solutions, Digital Video
Broadcasting methods and development.

Tautvydas Zelvys , M.Sc. of Computer Science.

Virtual Center of Excellence, Telecommunications department of Kaunas

150 www.erpublication.org