Professional Documents
Culture Documents
Efficient Multimedia Content Storage and Allocation in Multidimensional Cloud Computing Resources
Efficient Multimedia Content Storage and Allocation in Multidimensional Cloud Computing Resources
1/2, 2019
1 Introduction
Cloud computing had achieved various growth in the past years. In cloud data centres,
shared pools of servers are handled in order to yield on-demand computation,
communication and storage resources in the form of accessible services. Owing to the
elastic and on-demand characteristics of provisioning of resources, cloud computing
has been employed in multimedia applications increasingly in order to resourcefully
deal with the critical demands of resource. In cloud-based multimedia applications,
application providers rent a particular number of virtual machines (VMs) for delivering
multimedia services to users. For application providers, one important concern is about
the optimal deployment of multimedia tasks to different VMs for the purpose of
distributed processing. In order to deal with this issue, efficient scheduling schemes are
necessary.
Usually, cloud computing adapts two-level scheduling procedure. The first level
corresponds to the user-level scheduling, where the requests of users for one application
are distributed to various VMs based on the present workload. By means of balancing the
workload among VMs, the user-level scheduling can prevent the congestions in cloud.
The task-level scheduling is carried out in a finer granularity than user level scheduling.
Generally, a multimedia application can be divided into a group of tasks. Few tasks can
be executed in parallel, whereas few tasks have to be processed serially. The aim of
task-level scheduling is the assignment of tasks to VMs such that the whole execution
time can be reduced. The two-level scheduling is carried out in a hierarchical manner in
cloud. The optimal user-level scheduling has been explored in the earlier research work
(Nan et al., 2013). In this research work, the optimisation of task-level scheduling is
focused for cloud-based multimedia applications.
In cloud, efficient task-level scheduling can make the optimal match of the demands
of tasks with the resource provisioning of VMs and enhance the quality of service (QoS).
Nonetheless, it is a challenge to attain the optimal task-level scheduling. At first, there are
different precedence constraints that are the precedence relations among tasks. While
allocating the tasks to VMs, it is hard to obey all these precedence constraints. Secondly,
multimedia applications have diverse operation structures. Usually, there are three
fundamental structures: sequential, parallel and mixed structures.
In the sequential structure, tasks are run serially, with one task beginning just after
one earlier task has finished. In the parallel structure, tasks can be carried out in a
concurrent manner. Mixed structure integrates both sequential and parallel structures. It is
a challenging task to get an efficient scheduling scheme for various structures. Thirdly,
VMs have diverse resource capabilities. It is again a challenge to allocate a task to the
VM that suits the best. At last, application providers need an effective scheduling scheme
22 E. Sivaraman and R. Manickachezian
that can adapt to time-dependent constraints. In order to deal with the challenges that
are mentioned above, an optimal task-level scheduling is proposed for cloud-based
multimedia applications in this research work.
2 Related works
cloud resources in the absence of centralised management control. Shi and Zhan (2009)
focus on the maximisation of the steady-state throughput by deploying resources for the
independent equalised tasks in the cloud. Teng and Magoules (2010) present a resource
pricing and equilibrium allocation policy based on the consideration of cloud users’
competition for limited resources.
Handling multimedia content is the more tedious task in cloud computing environment
in which multimedia resources consume more amounts of resources with diverse
characteristics. Getting and scheduling the resources matching the requirements of
multimedia contents would be a hard task. In the research work, a new framework is
built, referred to as improved storage and scheduling of multimedia contents in cloud
storage (ISS-MCCS) that tries to yield the services to the users with the concern of
reliable and efficient scheduling. Multimedia contents would take up more amount of
storage that would be changed dynamically during run time. These problems are solved
in this research work by balancing the load level of resources during run time. The steps,
which are adopted for handling the multimedia contents effectively, are given below:
1 Irregularity in the multidimensional resource usage of a server for fuzzy neural
network (FNN).
2 Hybrid genetic-cuckoo search algorithm (HGCSA) that optimises the CMSdynMLB.
3 Hybrid fuzzy weighting techniques for QoS.
4 A second vital factor is the determination of an efficient caching machine to increase
the lifetime of the cached resource which is done by using adaptive replacement
cache (ARC). The concept in ARC is the replacement of the least often and least
currently utilised cached data.
In Section 3.1, an elaborate discussion over the newly introduced research methodology
is provided with appropriate diagrams and examples.
Figure 1 Illustration of multimedia cloud computing system (see online version for colours)
range values are then converted into their actual values. In the neural network, in case the
resources do not get mapped with the server clusters, back propagation algorithm
is employed, and the resources get reclassified and similar operations are followed.
The centralised database is exploited for the maintenance and updating of the training
data needed for neural networks. The feedback or feed forward algorithm is utilised for
the reclassification of resources in case it is not matched with the server cluster.
Figure 3 Time graph for all jobs (see online version for colours)
26 E. Sivaraman and R. Manickachezian
Model 1: u ( Ai ) =
∫u . μ ( u ) du
Ai
∫μ ( u ) du
Ai
Model 2: u ( Ai ) =
∫v ( A ) − u
i min
∫u − u
max min
where
v ( Ai ) =
∫u.μ ( u ) du
Ai
∫μ ( u ) du
Ai
Sup S
umax =
u
Inf S
umin =
u
Efficient multimedia content storage 27
n
S = ∪ Si
i =1
{
Si = u | µAi ( u ) > 0 }
Then, the grade of membership µAi of Ai for every model in the fuzzy numeric rating is
represented in equation (1).
Sup
μA = μA (u ) (1)
i
u i
Therefore, once the transformation is done, the fuzzy linguistic rating Ai would be a
fuzzy singleton having only one point in the fuzzy set. The fuzzy singleton of the
transformation of the fuzzy linguistic rating, Ai, can be defined in equation (2).
μA
Ai = i
(2)
u ( Ai )
As the fuzzy linguistic ratings or the fuzzy numbers that are considered are supposed to
be convex and normal, Ai can be indicated as Ai = 1/u (Ai). Therefore, the weighted-
average operation in fuzzy multi criteria decision making can be specified as below.
Let B1, B2, …, Bm refers to the alternatives to be evaluated by the criteria
C1, C2, …, Cn. Then, for a particular alternative Bi, the relative merit of criterion Cj is
evaluated by a fuzzy linguistic rating rij. Moreover, the relative significance of every
criterion is evaluated by means of a weighting coefficient wj for criterion Cj. Then, the
alternative Bi will get the weighted-average rating ri, expressed in equation (3).
∑
n
j =1
W j rij
r = (3)
∑
i n
W
j =1 j
where wj and rij are denoted by fuzzy numbers, wj and rij can be converted into wj and rij.
By Definition 1, and the transformed weighted average ri is expressed in equation (4)
∑
n
j =1
W j rij
r =
∑
i n
j =1
Wj
Hybrid cuckoo algorithm has both the benefits of genetic algorithm and cuckoo search
algorithm. Cuckoo search is an optimisation algorithm (finding a value x so that f(x) is
much as small (or as large) as possible) depending on the cuckoo species that lays its
eggs in the nests of other host birds (of other species). A GA is a random methodology
that is dependent on the principles of natural evolution and genetics. GA integrates the
exploitation of the earlier results with finding new areas of the search space. For the
purpose of overcoming the demerits of both cuckoo algorithm and GA, the new algorithm
referred to as hybrid cuckoo algorithm is realised.
3.5 Better cache management with the concern of optimal storage and
computation overhead
ARC is a typical page replacement algorithm having a better performance compared with
least recently used (LRU) designed at the IBM Almaden Research Center.
This is achieved by keeping track of both of the pages that are frequently used
and recently used in addition to a current eviction history for both. In 2006, IBM was
provided with a patent for the ARC policy. ARC replacement policy was utilised
in this research work, so as to yield a direct sequential read for one-time-only. This is
inclusive of the processing of the stored images so that they might pass through the cache
without any impact from the external events over the temporal locality. The ARC
replacement algorithm was also supposed to deal with longer periods of VM low
temporal locality by maintaining the VMs in the cache simultaneously, which can more
possibly is requested by the server again. Keeping this in consideration, it can be
ascertained that ARC replacement functions as a filter, efficient in differentiating and
tracking the temporal locality of VMs. In the event during the imposition of a random
load in the network, ARC will decide its source and then have itself reconfigured for
exploiting the chance.
In this research work, it was supposed that the optimal conditions for the operation of
the node might be dependent on the received request vector r at the maximum average
request rate. In case FC(r) presents the function returning the greatest values from r, C,
the cache probability vector for node k, Ck, is as illustrated in equation (5).
CkLFU = 1i∈FC ( rk ) (5)
where 1 indicates the indicator function. In the condition of the average VMs’ request
rates being in line with the ARC static, the images with higher average request rates in
cache VMs are instead utilised.
4 Experimental results
It is apparent from the earlier section that multiple researchers have developed
multimedia computing load balancing methods in distributed networks for resolving
the load imbalance problem. In this research work, an ISS-MCCS is proposed for the
optimisation of load in the MCC to distribute over servers with reduced cost for the
transmission and also the maximum usage of resource. The important feature
30 E. Sivaraman and R. Manickachezian
from existing paper is the support provided to the load in the dynamic environment
of the MCC. Simulation result shows that the algorithm proposed performs better in
ISS-MCCS.
The parameters below are utilised for the evaluation and after the evaluation of these
parameters, it can be said that the proposed research work performs better compared with
the earlier modelling.
1 Time
2 Energy consumption
3 Resource utilisation.
Three values [–1, 0, 1] are considered. In case the value is –1 then the resources are
utilised less, and if the value is 0 the resources are utilised properly and in case the value
is 1 then the resources are over utilised. Resources are correctly utilised in ISS-MCCS for
50 and 90 number of jobs correspondingly but in GA and cuckoo algorithm resources
are either less or over utilised. Therefore, it can be said that ISS-MCCS maximises the
resource utilisation as illustrated in Figures 5 and 6.
Figure 4 Energy for all jobs (see online version for colours)
Figure 5 Resource utilisation for 50 jobs (see online version for colours)
32 E. Sivaraman and R. Manickachezian
Figure 6 Resource utilisation for 90 jobs (see online version for colours)
5 Conclusion
In the case of MCC, resource management and provisioning is a complicated task that
cannot be carried out with accuracy along with enhanced user satisfaction level. In this
research work, a new framework referred to as ISS-MCCS is deployed, focused on the
improvement of efficiency and accuracy in the handling and management of resources for
the multimedia contents. This is accomplished by the right amount of load balancing
making use of FNN that dynamically varies the resource placement depending on the
server cluster capacity degree. Also the incoming user requests are scheduled optimally
considering the different QoS parameter employing fuzzy weighted multi-criteria
decision making and hybrid genetic cuckoo search algorithm. And at last, the content
caching is optimised based on the LRU concept for the purpose of reducing the storage
and computational overhead employing ARC mechanism. The experimental tests
performed prove that the proposed research work yields better result in comparison with
the available research techniques.
References
Aljaber, B., Jacobs, T., Nadiminti, K. and Buyya, R. (2007) ‘Multimedia on global grids: a case
study in distributed ray tracing’, Malaysian Journal of Computer Science, Vol. 20, No. 1,
pp.1–11.
Balasangameshwara, J. and Raju, N. (2012) ‘A hybrid policy for fault tolerant load balancing in
grid computing environments’, Journal of Network and Computer Applications, Vol. 35,
No. 1, pp.412–422.
Chen, D. and Zhao, H. (2012) ‘Data security and privacy protection issues in cloud computing’,
Proceedings of International Conference on Computer Science and Electronics Engineering,
Vol. 1, pp.647–651.
Foster, I., Zhao, Y., Raicu, I., and Lu, S. (2008) ‘Cloud computing and grid computing 360-degree
compared’, Grid Computing Environments Workshop, pp.1–10.
Efficient multimedia content storage 33
Fusenig, V. and Sharma, A. (2012) ‘Security architecture for cloud networking’, Proceedings of
International Conference on Computing, Networking and Communications, pp.45–49.
Gitinavard, H., Mousavi, S.M. and Vahdani, B. (2016) ‘A new multi-criteria weighting and ranking
model for group decision-making analysis based on interval-valued hesitant fuzzy sets to
selection problems’, Neural Computing and Applications, Vol. 27, No. 6, pp.1593–1605.
Kaveh, A. and Bakhshpoori, T. (2016) ‘An efficient multi-objective cuckoo search algorithm for
design optimization’, Advances in Computational Design, Vol. 1, No. 1, pp.87–103.
Lin, W. and Qi, D. (2010) ‘Research on resource self-organizing model for cloud computing’,
Proceedings of International Conference on Internet Technology and Applications, pp.1–5.
Muñoz, P., Barco, R. and De la Bandera, I. (2013) ‘Optimization of load balancing using fuzzy
Q-Learning for next generation wireless networks’, Journal Proceedings of Expert Systems
with Applications, Vol. 40, No. 4, pp.984–994.
Nan, X., He, Y. and Guan, L. (2011) ‘Optimal resource allocation for multimedia cloud based on
queuing model’, Proceedings of International Workshop on Multimedia Signal Processing,
pp.1–6.
Nan, X., He, Y., and Guan, L. (2013) ‘Optimization of workload scheduling for multimedia cloud
computing’, International Symposium on Circuits and Systems (ISCAS), pp.2872–2875.
Shi, H. and Zhan, Z. (2009) ‘An optimal infrastructure design method of cloud computing services
from the BDIM perspective’, Proceedings of Asia-Pacific Conference on Computational
Intelligence and Industrial Applications, Vol. 1, pp.393–396.
Silberstein, M., Geiger, D., Schuster, A., and Livny, M. (2006) ‘Scheduling mixed workloads in
multi-grids: the grid execution hierarchy’, Proceedings of International Symposium on High
Performance Distributed Computing, pp.291–302.
Tai, J., Zhang, J., Li, J., Meleis, W., and Mi, N. (2011) ‘Ara: adaptive resource allocation for cloud
computing environments under bursty workloads’, Proceedings of International Conference
on Performance Computing and Communications, pp.1–8.
Tan, M., Siegel, H.J., Antonio, J.K. and Li, Y.A. (1997) ‘Minimizing the application execution
time through scheduling of subtasks and communication traffic in a heterogeneous computing
system’, IEEE Transactions on Parallel and Distributed Systems, Vol. 8, No. 8, pp.857–871.
Teng, F. and Magoules, F. (2010) ‘Resource pricing and equilibrium allocation policy in cloud
computing’, Proceedings of International Conference on Computer and Information
Technology, pp.195–202.
Wang, Y., Zhou, Z., Liu, L. and Wu, W. (2013) ‘Replica-aided load balancing in overlay
networks’, Journal of Network and Computer Applications, Vol. 36, No. 1, pp.388–401.
Wu, Y., Wu, C., Li, B., Qiu, X. and Lau, F. C. (2011) ‘Cloudmedia: when cloud on demand meets
video on demand’, Proceedings of International Conference on Distributed Computing
Systems, pp.268–277.
Xiu, J., He, Q., Yang, Z., and Liu, C. (2016) ‘Research on a multi-objective constrained
optimization evolutionary algorithm’, Proceedings of International Conference on Cloud
Computing and Intelligence Systems, pp.282–286.
Yu, J. and Buyya, R. (2005) ‘A taxonomy of scientific workflow systems for grid computing’,
ACM SIGMOD Record, Vol. 34, No. 3, pp.44–49.
Zhu, W., Luo, C., Wang, J. and Li, S. (2011) ‘Multimedia cloud computing’, IEEE Signal
Processing Magazine, Vol. 28, No. 3, pp.59–69.