06-05-2013, 04:00 PM
Wireless Network Aware Cloud Scheduler for Scalable Cloud Mobile Gaming
Wireless Network Aware.pdf (Size: 693.68 KB / Downloads: 28)
Abstract
Cloud Mobile Gaming (CMG) [1][10][11] has been
proposed as an approach to enable rich Internet games on mobile
devices, where the rendering of the games is performed on cloud
servers, as opposed to on mobile devices. Though promising, the
CMG approach may require significant cloud computing resources
for the concurrent gaming sessions, and even more critically,
significant bandwidth for delivering the rendered videos back to
mobile devices, leading to high cloud costs, and questions regarding
resource constrained wireless networks. This paper addresses the
problem of making the CMG approach scalable and economically
feasible by proposing a novel Wireless Cloud Scheduler (WCS),
which can increase the number of simultaneous CMG sessions that
can be supported while ensuring Mobile Gaming User Experience
(MGUE)[1] with the available wireless network resources, while
minimizing the cloud service cost incurred by the CMG provider.
Unlike conventional network schedulers, WCS considers
simultaneously the constraints of the wireless networks that may be
available to each CMG user, including cellular and WiFi, as well as
the cost of available cloud resources, while scheduling the most
optimal wireless link and cloud server for each CMG session. To
further enhance the performance of WCS, we also propose a joint
scheduling-adaptation algorithm, that can systematically leverage
adaptation techniques introduced in [10][11] to adapt the
communication needs of in-service users if the available wireless
network bandwidth is not sufficient for a new CMG user. Our
simulation results demonstrate that the use of WCS, and the joint
scheduling-adaptation algorithm, can significantly improve the
performance of the CMG approach, increasing the number of
simultaneous CMG sessions that can be supported, while
maximizing aggregate MGUE and minimizing the average cloud
service cost.
INTRODUCTION
With the tremendous growth in the market of mobile devices,
there is a growing desire to enable 3D, multiplayer, Internet
video games on smart phones. However, it is difficult for thin
mobile devices to use the traditional gaming approach for rich
3D games, where the gaming clients execute the data,
computation and energy intensive tasks such as 3D graphic
rendering. Instead, a new Cloud Mobile Gaming (CMG)
approach [1][10][11], where the responsibility of executing the
gaming engines is put on CMG servers instead of the mobile
devices, has the potential for enabling mobile users to play the
same rich Internet games available to PC users.
While a CMG service can be launched in multiple ways, we
assume the architecture and eco-system shown in Figure 1. The
CMG provider “rents” cloud servers from cloud platform
providers to host the CMG engines that need to be executed for
each CMG session. We assume the mobile users have WiFi
and/or cellular access provided by their wireless network
provider, and paid by the mobile users according to the data rate
plans.
CHALLENGES AND OBJECTIVES OF WIRELESS CLOUD
SCHEDULER FOR CMG PROVIDER
In this section, we explain the challenges and summarize the
objectives of wireless cloud scheduler. The proposed WCS
should firstly be able to schedule each requesting CMG user to a
wireless network and a CMG server, so as to satisfy Quality of
Service (QoS) requirements for an acceptable MGUE.
According to our previous studies in [1], the major factors
affecting MGUE are game response time, and video data rate
associated with the quality of gaming video. For each game,
there is a certain Response Time Acceptable threshold RTA,
above which user cannot accept the gaming quality. And for
each CMG session, there is an associated computing resource
requirement CMGComp, a storage space requirement CMGStorage,
and a communication video data rate requirement CMGDataRate,
which are determined by the type of game played and the game
resolution used. Therefore, the QoS requirements of each CMG
session consist of RTA, CMGComp, CMGStorage, and CMGDataRate.
Second, though cloud computing resource is elastic and
unlimited, the bandwidth of wireless mobile network is limited.
It is possible that a user request cannot be scheduled due to the
network bandwidth constraint. Thus for CMG service, there is a
schedule rate, a term that denotes the percentage of the users
who are actually being served in CMG system (while satisfying
QoS requirements for each served user). For example, if the
number of requesting customers is 100 and we can only schedule
90 of them, then the schedule rate is 90%. The WCS should
achieve as high schedule rate as possible in a given network
bandwidth constraint.
SCHEDULING ALGORITHM FOR WCS
Traditional network and processor scheduler usually
calculates the optimal solution by looking into the entire system
resources and considering all the users/tasks, including new
requesting users/tasks and in-service users/tasks. This optimal
solution will very likely have to reschedule in-service users/tasks
to the new resources. However, such scheduling approach may
not be appropriate for CMG application. Firstly, because the
CMG scheduler will need to cater to a very large number of
CMG sessions using very distributed wireless network and cloud
computing resources (as opposed to a base station scheduler
caring about a few sectors, or a processor scheduler caring about
a few cores or processors), it may be computation and time
consuming to recalculate the optimal schedule every time a new
CMG request comes. And more importantly, once a CMG
session starts, it is very diffcult to reschedule the network access
method and cloud server for this CMG session. Therefore, in this
paper we will propose an in-service dynamic scheduling
algorithm, which looks into the available resources to make the
scheduling decision for the new users only.
Results and Conclusions
The results of the simulation experiments are shown in figure
4. Figure 4(a) presents the relationship between the schedule rate
(the percentage of simultaneous CMG sessions that can be
scheduled) and the number of users that has requested CMG
service. From the results in figure 4(a), we note that WCS may
not be able to schedule some of the incoming users when we
keep randomly adding users into the CMG system. This is
mainly because some requesting users have entered into the
regions where network bandwidth is fully utilized. It can be also
observed from figure 4(a) that the schedule rate if using F2(m,n)
or F3(m,n) is always better than using F1(m,n). This is because
F1(m,n) (MGUE-based utility function) has not considered the
utilization of wireless networks. For example in figure 3, with
utility function F1(m,n), every user who has access to both WiFi
and Cellular network will be scheduled to use Cellular network
since it offers smaller network delay and therefore better gaming
experience. Cellular networks will be quickly fully utilized, so
that the next coming users, who only have access to Cellular
networks, will not be able to be scheduled.