02-02-2013, 10:41 AM
Game Theory Based Dynamic Bit-Rate Adaptation for H.264 Scalable Video Transmission in 4G Wireless Systems
1Game Theory Based Dynamic.pdf (Size: 117.74 KB / Downloads: 40)
Abstract
In this paper, we propose a game theoretic framework
for decentralized H.264 scalable video bitrate adaptation
in 4G wireless networks. The framework presented employs a
pricing based utility function towards video streaming quality
optimization. An algorithm is presented for iterative strategy
update of the competing scalable coded video streaming end users
towards an equilibrium allocation. In this context we demonstrate
the existence of a Nash equilibrium for the proposed video bitrate
adaptation game based on the quasi-concavity of the net video
utility function. Existence of Nash equilibrium ensures efficient
usage of the 3G/4G bandwidth resources towards video quality
and revenue maximization. We present simulation results to
comprehensively illustrate the performance of the above game
theoretic video quality adaptation scheme. We also discuss the
behavior of the parameters of the net video utility function,
their impact on video bit-rate allocation and overall bandwidth
utilization of the wireless network.
INTRODUCTION
Next generation 4G wireless networks such as WiMAX [1],
LTE [2] are envisaged to provide a plethora of multimedia
content based rich services to the users. The demand for
video based services such as video conferencing, interactive
gaming and subscription based broadcast/ multicast services is
progressively increasing in next generation wireless networks.
Thus, reliable video transmission holds the key to efficiency
and performance in 4G cellular networks. Towards this end,
4G systems often impose Quality of Service (QoS) restrictions
on content delivery to ascertain and satisfy a minimum level
of performance for the subscribers. Moreover, the disparate
link qualities, display resolutions and device capabilities of the
cellular users cannot be handled by conventional static fixed
profile video codecs. In this context, H.264 based scalable
video transmission provides the flexibility of coded video bitstream
rate adaptation. It is therefore essential to develop an
appropriate framework to optimally allocate the scalable video
streams to different users/ groups in the wireless network for
maximization of bandwidth efficiency. The standard scheduling
services such as rtPS and nrtPS available in 4G WiMAX
are not targeted towards overall video quality maximization,
especially in scalable video scenarios.
SYSTEM MODEL
In this section we describe the standard parametric models
to characterize the rate and quality functions of an H.264
coded scalable video stream. The quantization parameter qi
which corresponds to SNR scalability decides the bitrate and
the quality of the video sequence. A reduction in qi will result
in higher bitrate/ quality and an increase in qi will result in
lower bitrate/ quality of the video sequence. The bitrate Ri (qi)
of the scalable video stream of the ith user at a fixed frame
rate t and quantization parameter qi is given from [3].
GAME THEORY FRAMEWORK
We present below a brief overview of the basic concepts
and notation employed in the context of game theory [11]. A
game is essentially a framework to characterize the competitive
interaction between N independent decision making agents
termed as players for allocation of a limited set of resources.
These interactions are governed by a set of rules. For instance,
in the context of a wireless network, the players are the users
competing for allocation of time-frequency resources and the
governing rules are those dictated by the regulating agencies
such as the FCC. Each player i, 1 ≤ i ≤ N in a game has a set
of strategies denoted by Si from which he chooses a strategy
si ∈ Si. Let s−i denote the vector of strategies of all the
players except the ith player. Each player/ user is associated
with a utility function ui which denotes his reward ui (si, s−i)
corresponding to the outcome of the game.
CONCLUSION
In this paper we proposed a pricing based game theoretic
scheme for decentralized resource allocation towards H.264
based scalable video transmission in 4G wireless networks.
This novel competitive approach for dynamic video resource
allocation is based on a joint video quality and parametric
pricing based quasi-concave net-utility function for video
transmission. Employing the framework of game theory it was
demonstrated that such a scheme guarantees the existence of
the Nash equilibrium based stable video bit-rate allocation in
which no player has an incentive to deviate. A competitive
video bit-rate adaptation procedure was proposed based on
an iterative strategy update procedure through MAC based
polling available in the 4G WiMAX standard.