Adaptive Optimization of Cloud Security Resource Dispatching SFLA Algorithm

The International Journal Of Engineering And Science (IJES)
|| Volume || 4 || Issue || 3 || Pages || PP.39-43|| 2015 ||
ISSN (e): 2319 – 1813 ISSN (p): 2319 – 1805
www.theijes.com The IJES Page 39
Adaptive Optimization of Cloud Security Resource Dispatching
SFLAAlgorithm
1,
Xiaoxi Hu
1,
School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing,
100083, China
---------------------------------------------------------ABSTRACT----------------------------------------------------------
According to the present problem of cloud computing resources dispatching model as low safety performance, a
cloud computing resources safety dispatching model is proposed based on the adaptive optimization SFLA
algorithm. First of all to adaptive optimize the learning factor of standard SFLA algorithm, at the same time of
the continuation of the last update inertial step, also to learn to the history optimal value of individual
neighborhood in the memory, expand the search area of the individual, maintaining the diversity of the
population, and as the original algorithm is easy to fall into local optimal problem, the Cauchy mutation operator
is introduced into the basic SFLA algorithm, and improve the locally adaptive update strategy. Simulation
experiments show that the proposed cloud computing resources security dispatching model has better
convergence effect and application effect based on adaptive optimization SFLA algorithm.
Keywords - SFLA, Adaptive optimization, Cloud computing resource dispatching, Security dispatching, Local
update strategy, Learning factor
--------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: 09 March 2015 Date of Accepted: 20 March. 2015
--------------------------------------------------------------------------------------------------------------------------------------
I. INTRODUCTION
With the rapid development of cloud computing, cloud amount and internal server number of each cloud
platform are also in rapid growth, the operation of those large number of server cluster depends on the cloud
computing system. Cloud computing users access to information resource service through the network
conveniently on-demand as easily as get water and electricity; On the other hand, cloud computing can compatible
with a variety of based resources of software and hardware, enables the different resources to division of labor to
work together, to implement the dynamic flow of resources; The two basic function is the core of the cloud
computing system. Thus, resource scheduling strategy in cloud computing system for the operation of cloud
platform is particularly important, it directly affect the speed of cloud users experience, cloud platform equipment
utilization, cloud service cost and economic benefit etc.
Scholars at home and abroad research on safety scheduling of cloud computing resources is essentially
revolve around the following aspects. Amazon provides the cloud IaaS (infrastructure as a service) products,
provides computing power, storage, and other services for users, according to Amazon, its Web service allows
you to use the global computing infrastructure of Amazon.com, this is Amazon’s retail business and the core of
corporate affairs. Trusted computing technology is used to protect the cloud computing environment, the premise
of set up the scheme is to establish a trusted computing base, and then use the trusted computing base to control
and manage the upper service software, Santos et al. propose a trusted cloud computing platform TCCP, based on
the platform, IaaS cloud computing service providers can provide users with a closed box execution environment,
to ensure that the user virtual chance the confidentiality. In addition, it also allows users to inspect IaaS cloud
computing service provider’s service is safe or not before starting virtual machine. Companies such as EMC, Intel
and VMware announce the cooperation project of “trusted cloud architectures”, and put forward a proof of
concept system. Boneh put forward a kind of support other authorized user retrieval scheme, cipher processing
focuses on privacy homomorphism encryption algorithm design. Mowbray et al. propose a privacy management
tool on client bases, and put forward the trust model with user-centric, to help users to control their sensitive data
in the cloud storage and use. Rankova et al. propose an anonymous data search engine, can make interactive
search on each other’s data, to obtain data information what you want, at the same time ensure that the content of
search inquiry will not be known by each other, search with request irrelevant information will not be obtained. In
cloud computing environment, when sharing and protecting the resources, must formulate a reasonable and
mutually acceptable access control policy for shared resources. Therefore, we need to compound support strategy.
Mclean first raised the issue in mandatory access control framework, he compound two safety synthesis to a new
lattice structure, constructs a synthetic framework of mandatory access control strategy. Synthesis strategy should

Adaptive Optimization of Cloud Security Resource Dispatching SFLA Algorithm
not only guarantee the safety of the new strategy, but also ensure that new synthetic strategies cannot contrary to
the original access control policy of each domain. Therefore, Gong put forward the principles of autonomy and
safety. Bonatti presents a synthetic algebra of access control policy, based on set theory using synthetic operator
for synthesis of security policy. Wijesekera put forward strategy based on authorized state change to composite
algebraic framework, Agarwal structure strategy synthetic solution of semantic Web services, Shafiq put forward
for more than a trust domain RBAC synthesis strategy, focus on strategies in the synthesis of solution and the
domain the consistency problem of original strategy.
According to the characteristics of cloud computing, a cloud computing resources safety dispatching
model based on the adaptive optimization SFLA algorithm is proposed, in order to improve the security of cloud
computing resources dispatching.
II. DISPATCHING SFLA ALGORITHM OF RESOURCES SECURITY
Hybrid leapfrog algorithm (SFLA) algorithm is a bionic optimization algorithm of global collaborative
search, the algorithm is inspired by the frog foraging behavior: a group of frogs living where there are many stones,
frog by jumping between stones and looking for food. Each frog looks for different stones to improve their ability
to search for food, and communicate ideas with each other between frogs to achieve information sharing.
For optimization problem of M dimensional space, suppose that the frog population size as S , frog i marked
1 2( ) { , ,..., ,..., }i i ij iMU i U U U U as a solution of solution space, 1 i S  , 1 j M  . Compute fitness
value of each frog within the population, descending order according to the fitness value. The entire frog
populations were divided into m groups, each group contains n frogs, which meet the relationship S m n  .
At the same time, according to the following rules to sort frog population: first frog into the first group, second
frog into 2 groups, and the rest can be done in the same manner, the m frog into the m groups. And then, the
1m  frog into the first group, the 2m  frog into second groups, such as the cycle distribution, until finish all
the frogs were assigned.
The best and the worst of the fitness value of the frog in ethnic groups, respectively as bP and bP . The
best fitness value of frogs in the population as gP . Before reach the preset local the number of iterations, local
search in the group of all nationalities, through the study manner of wP in ethnic groups to the groups bP , to
achieve the optimization for wP . After local search in each group, the frog jumps between ethnic groups to mix
into a population, so that complete a global search.
Shuffled Frog Leaping Algorithm (SFLA) can be basic divided into three processes as global search, local search
and mixed operation. The global search process is as follows:
1) Algorithm parameters. Number of ethnic groups m , local search the maximum number of iterations
maxJ , global maximum number of iterations maxQ , each group contained in the number of frogs n , The frog
population size S m n  , the dimensions of the optimization problem solution space M , and upper and lower
limits of search space respectively marked as: H , L ;
2) Initialize the population. In M dimension the feasible solution space, randomly generated S frogs
marked as (1), (2),..., ( )U U U S , the i frog marked as 1 2( ) { , ,..., ,..., }i i ij iMU i U U U U , 1 i S  , 1 j M  ,
[ , ]ijU L H ;
3) Fitness function. Through fitness function ( )f i to evaluate performance of each frog solution ( )U i ;
4) The frog population division. Frog rank: according to fitness function ( )f i , the S frogs descending
order according to the fitness, generates an array { ( ), ( ), 1,2,..., }P U i f i i S  , the frog (1)U is the best individual
in the frog populations as global extreme value 11 12 1 1(1) { , ,..., ,..., }g j MP U U U U U  ; The frog divided race:
according to the equation (1) to divide the frog population.
( ) ( ( 1))kZ i U k m i    (1)
In equation, 1,2,...,i n , 1,2,...,k m , distributing the sorted frogs into m groups 1 2, ,..., mZ Z Z , each group
contained n frogs; for example 3m  , 2n  , that is 6S  , the frog (1)U into (1)U groups, (2)U enter into
2Z , (3)U enter into 3Z , (4)U enter into 1Z , (5)U enter into 2Z ,U (6) enter into 3Z .
Perform evolution within each group. Within the group, the frog is affected by the other frogs’ individuals culture
within the same ethnic, through evolution, to make each frog learning approach to extremum in their ethnic local.

III. ADAPTIVE OPTIMIZATION OF SFLA ALGORITHM
3.1 Adaptive Optimization of Learning Factors
Because the frog individuals with memory function, can remember the last update step length and
its own history optimal value of individual neighborhood, and so wP learn not only from bP or gP in
iteration process, at the same time of the continuation of the last update inertial step, also to learn the
history optimal value from individual neighborhood in the memory, with the increase of the number of
iterations, learning factors impact on individual update strategy is linear weakening trend. While this
update strategy speed up the convergence, expand the individual search area, to maintain the population
diversity, a certain extent, expanded the search ability of the algorithm. Specific update strategy such as
equations (2) ~ (4).
1 2
3
( 1) ( ( ) ( ))
( )
w
b w
Dis t W R Dis t R His P
R P P
  
 
(2)
max
max
( )e s e
T t
W W W W
T

   (3)
( 1) ( ) ( 1)w wP t P t Dis t    (4)
In equation, 1R , 2R , 3R are a random number between 0 and 1, Dis is moving step of wP , t is the number
of iterations, maxT is the total number of iterations, W is a weighting factor, sW and eW is the initial value
and end value weighting factor, in this paper, the values are respectively 0.9 and 0.4, ( )wHis P is individual
neighborhood history optimal value of wP .
3.2 Adaptive Optimization of Local Update Strategy
In basic SFLA, each iteration only to update operation of ethnic group wX , and each use equation
(1) update the frog individuals. If the frog ignore some search domain, the algorithm is easy to fall into
local optimum. Aiming at this problem, to make full use of information among groups, SFLA algorithm
was proposed based on disturbance Cauchy distribution.
Cauchy distribution, is very important application of probability distribution to many areas such as a
mathematical theory, engineering application. Cauchy density function is defined as:
2 2
( )
[ ( ) ]
f x
x

  

 
(5)
Usually express by ( , )C   , its Cauchy probability distribution function is:
arctan(( ) / )
( ) 0.5
x
F x
 


  (6)
When 0  , 1  , known as the standard Cauchy distribution, to remember as (0,1)C . Cauchy
distribution is the shape of two relatively flat, wide and pattern is a located above the horizontal axis of
peak type curve (as shown in figure 1).
Fig.1: Cauchy Distribution Function

The Cauchy mutation operator is introduced into the basic SFLA algorithm, the local update strategy was
improved, the worst individual wX to subgroup within individual bX according to the equation (7) the
best learning update, the update strategy for:
( 1) ( ) (0,1)b wD t rand X X C     (7)
Compared to Gauss mutation, Cauchy mutation with larger probability, in a broader space domain, finds
the best solution. By adding random disturbance, improve diversity of the group, the lower the possibility
of the frog “into” local optimal domain, improve local escape ability of the algorithm. Compared with
Gaussian mutation, although Cauchy mutation are more likely to escape from the local optimal area, when
the algorithm into the late search, near the global optimal extremum, because of its high Cauchy mutation
operator variable step length easily over a better search area, is not conducive to the depth of local search.
To reduce the dramatic differences between the individual spaces by local update operation, put forward a
kind of Cauchy mutation hybrid leapfrog algorithm based on probability of disturbance. Local update
strategy is as follows:
( ) (0,1),
( 1)
( ),
b w
b w
rand X X C R Q
D t
rand X X R Q
   
  
  
(8)
( 1) ( ) ( 1)w wX t X t D t    (9)
In equation, R is random number rand of [0,1] , Q probability of disturbance.
IV. THE SIMULATION OF ALGORITHM
In order to verify the effectiveness of the improved algorithm proposed in this paper, simulation
experiments on it. First Sphere function 1f , Rosenbrock function 2f and Rastrigin function 3f , Griewank
function 4f and Schaffer function 5f were introduced in standard SFLA algorithm and the improved SFLA
algorithm for testing. The results are shown below:
Tab.1: The Test Results of Standard Algorithms and Improved Algorithm
f Algorithm Average optimal value Standard deviation Running time
1f SFLA 0.0062 0.0047 0.9
IM-SFLA 0.0028 0.0001 1.4
2f SFLA 124.9740 53.1455 1.0
IM-SFLA 45.9851 27.2636 1.7
3f SFLA 11.4995 3.6283 1.4
IM-SFLA 10.2293 4.3324 1.9
4f SFLA 0.5815 0.2399 1.5
IM-SFLA 0.1536 0.0805 1.9
5f SFLA 19.6511 5.1459 3.1
IM-SFLA 19.3725 7.8156 3.7
Seen from the data in table 1, under the same number of iterations and precision, based on the
improved SFLA algorithm can obviously get the better average optimal value than the standard hybrid
leapfrog algorithm, and its standard deviation is smaller, means that improved leapfrog algorithm has
better stability.
Then the standard SFLA algorithm and the improved SFLA algorithm in cloud computing
resources safety scheduling optimization, test its convergence, the result is as follows:

Fig.2: Convergence Test Results of Standard algorithms and Improved algorithm
Seen from the simulation results, the proposed improved SFLA algorithm has good convergence, and has
good application in cloud computing resources safety dispatching optimization.
REFERENCES
[1] Zhang Yan. A Novel Improved Bat Algorithm for Cloud Computing Resources Scheduling Optimization, Bulletin of Science and
Technology. 2014, 30: 117-121.
[2] Yin Fan. Algorithm of Multi-Agent Genetic Based on Cloud Computing Resource Scheduling. Bulletin of Science and Technology.
2014, 30: 247-249.
[3] Lv Shuhong. Research on the Cloud Computing Resource Scheduling Based on Chaos Strategy and Bat Optimization Algorithm.
Bulletin of Science and Technology. 2014, 30: 144-149.
[4] YE Feng. Application of Two-Way Convergence Ant Colony Algorithm in QoS of Cloud Computing Resource Scheduling.
Electronics Optics ＆ Control. 2014, 21: 93-96.
[5] Zhang Hengwei. Cloud resource scheduling method based on estimation of distribution shuffled frog leaping algorithm.
Application Research of Computers. 2014, 31: 3225-3228.
[6] Ji Keke. Application of Improved SFLA Algorithm in Resource Scheduling of Cloud Computing, Bulletin of Science and
Technology. 2014, 4: 225-228.
[7] Xu Bingyuan. Virtual Resources Dispatching Algorithm Based on Multipath Traversal Priority Selection in Cloud Computing,
Electric Power. 2014,8: 139-143.
[8] Feng Guofu. Maximum Profit Oriented Adaptive Scheduling of Cloud Resources, Mini-micro Systems. 2014, 35: 1717-1721.
[9] Ma Junbo. Security-constrained workflow scheduling in cloud computing environments, Computer Engineering ＆ Science. 2014,
36: 607-614.
[10] Luo Liang. Energy Modeling Based on Cloud Data Center, Journal of Software. 2014, 25: 1371-1387.
[11] Xu Bo. Virtual Machine Resource Scheduling Multi-objective Optimization in Cloud Computing, Journal of System Simulation.
2014, 26: 592-595.
[12] Hao Liang. A cost constrained resource scheduling optimization algorithm for reduction of energy consumption in cloud
computing, High Technology Letters. 2014, 5: 458-464.

Adaptive Optimization of Cloud Security Resource Dispatching SFLA Algorithm

More Related Content

What's hot(9)

Viewers also liked(19)

Similar to Adaptive Optimization of Cloud Security Resource Dispatching SFLA Algorithm(20)

Recently uploaded(20)

Adaptive Optimization of Cloud Security Resource Dispatching SFLA Algorithm