TITRE Adaptive Packet Video Streaming Over IP Networks - LaBRI

• We design an algorithm for fine grained TCP-Friendly video rate adaptation (see
Section 4.3).
Thus, this chapter presents our design philosophy in the transport of MPEG-4 video stream
over IP network. We enhance the transport layer by building several mechanisms and algorithms
that work together in order to achieve a seamless quality of service build around a cross-layer video
streaming system. Each mechanism is presented and evaluated both using network simulation and /
or experimental network. We finish this chapter by a conclusion which is presented in Section 4.4.
In order to understand our models and mechanisms, readers are invited to examine the
Appendix A which gives a consistent overview of the MPEG-4 standard.
4.1 A Content-Based Video Classification Model
To implements an efficient transmission of object-based MPEG-4 video over IP networks
with QoS management capabilities, the MPEG-4 Audio Visual Objects (AVOs) are classified based
on application-level QoS criteria and AVOs semantic descriptors according to AVOs descriptors
and MPEG-7 framework [152]. Thus, the classification model lead to a relative priority score (RPS)
for each AVO. The MPEG-4 AVOs requiring the same QoS performance (with the same RPS)
from the network are automatically classified and multiplexed within one of the IP Diffserv Per
Hop Behaviors (PHB). Object data-packets within the same class are then transmitted over the
selected transport layer with the corresponding bearer capability and priority level. Thus, we
propose to extend the MPEG-4 system architecture with a new “Media QoS Classification Layer”.
This layer implements the content-based video classification model. In our implementation, the
“Media QoS Classification Layer” makes use of a neural network classification model that is
transparent to the video application and the network layers.
Classification has been the subject of frequent and profound investigation. It has proved a
useful tool in real world applications. In networking environment, packet classification can be used
in network elements such as switch and router for packet switching, forwarding and filtering. The
proposal [149] presents an architecture that meets this goal. In [150], the authors present some
algorithms that can be used for packet classification and can be categorized as basic search
algorithms, geometric algorithms, heuristic algorithms, or hardware-specific search algorithms.
These algorithms are used in IP services such as firewalls and quality of service.
In the machine-learning mechanisms, classification methods are incredibly used. One wellestablished
approach is Bayesian classification, a technique that has become increasingly popular in
the recent years in part due to recent developments in learning with Bayesian belief networks [151].
Another classification method based on similarity such as K-NN (K-Nearest Neighbor), Naïve-
Bayes and hierarchical clustering are very used in machine learning.
In classification, neural networks solve many problems that conventional methods cannot, or
at least not within acceptable cost or performance criteria. In our design, we have used neural
network algorithms for automatic AVO classification.
64