International Journal of Application or Innovation in Engineering& Management (IJAIEM)

Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com

Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 34



ABSTRACT
Cloud Computing model, a large scale distributed computing infrastructure is built in order to meet the demand for power and
memory storage, increasing vigorously due to scientific research and industrialization. It is a recently proposed model
providing access to the distributed resources consistently. Maintaining the security, confidentiality and integrity of both users
and resources is the most challenging aspect of cloud. The main disturbing element of the security of the cloud i.e. the DDoS
attacks has led to the establishment of various technologies in order to gain defense against DDoS attacks. So here, this paper
will propose a defense mechanism against the DDoS attacks which is known as cloud specific Intrusion Detection System. This
defense mechanism will be able to detect the attack before the DDoS attack succeeds.
Keywords: Cloud, DoS, DDoS, Defense, Security.
1. INTRODUCTION
The main goal of cloud computing model is to provide the user an instant access of the resources they need in order to
support the scientific computation. Many researchers are looking forward to use the cloud approach for many different
applications. Since it is a recently proposed model, new to the technology world, managing security is still a major
issue. Although the cloud infrastructure ensures a number of security features, an intruder can still explore many
security flaws. If an intruder gets unauthorized access to the cloud, the cloud resources can be misused in many ways.
Among the numerous types of attacks, the intruder can also choose the most deadly attack i.e. distributed denial-of-
service attack. These attacks flood the system with an excessive amount of attack traffic thereby consuming bandwidth
and network resources. So, in order to achieve protection against such attacks this paper will establish a defense
mechanism fighting immediately against these attacks. So, in this paper, in order to achieve the goals of managing
security in the cloud computing model we will propose an intrusion detection system. This mechanism will prevent the
cloud user from the attack before the attack succeeds.
2. THE CLOUD AND THE SECURITY MEASURES
The cloud computing model is based on three tenants: infrastructure as a service (IaaS), platform as a service (PaaS)
and software as a service (SaaS)[1].
2.1 Infrastructure as a Service
This capability provides storage and many computing resources to the customer in order to deploy and run the software.
2.2 Platform as a Service
This capability provides applications involving programming languages, which are to be deployed onto the cloud
infrastructure.
2.3 Software as a Service
This capability helps to use the above mentioned applications running on the cloud infrastructure. Now, the four
deployment models for the cloud services are:
2.3.1 Public Cloud
It helps in making available the cloud infrastructure to the general public or a large industry group or an organization.
2.3.2 Private Cloud
This model enables the third party to make available cloud infrastructure to operate for a single organization.
2.3.3 Community Cloud
In this model, the cloud infrastructure supports a community that has shared concerns. It can also be managed by some
organization or the third party.
2.3.4 Hybrid Cloud
This model enables the cloud infrastructure established by two or more clouds to remain unique but bound together by a
standardized technology [2].
A Dual Mechanism for defeating DDoS Attacks
in Cloud Computing Model

Upma Goyal
1
, Gayatri Bhatti
2
and Sandeep Mehmi
3


1,2,3
Computer Science Department, Punjab Technical University, Kapurthala, India

International Journal of Application or Innovation in Engineering& Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 35

3. THE CLOUD COMPONENTS
Cloud comprises of five components [3]. These are as follows:
3.1 A Portal
It acts as a user interface through which we can login and use the cloud. After the login step, the user can submit a task.
3.2 A Service Broker
This identifies the resources involved in handling a particular task submitted by the user.
3.3 A Task Scheduler
This component of the cloud computing model sets the rules for scheduling the jobs on the cloud infrastructure.
3.4 A Task Manager
It launches the submitted task.
3.5 A Group of Cloud Nodes
These can be the servers, the clusters, the work stations belonging to the different LANs, WANs or the internet.
4. EXECUTING THE TASK IN THE CLOUD
In the cloud computing model [4], the execution of the task is carried out in four steps as shown in figure 1. These
are as follows:
1. The user logs in to the cloud through the portal and submits the task to the task manager after the verification.
2. The task manager checks the available resources and from them picks up a list of resources in order to satisfy that
particular task.
3. The scheduler sets the rules on the cloud infrastructure in order to schedule the submitted task.
4. The job manager supplies data to the listed resources and returns the computed result to the user after the completion
of the task.

Figure 1 Executing the Task in the Cloud


5. THE CLOUD ARCHITECTURE
Cloud has a four layered architecture [5] which identifies the systems fundamental components and specifies the
function of these components as shown in figure 2. These four layers are:
5.1Cloud Fabric
This layer comprises of all the resources accessible from the internet. It also includes the protocols and the interfaces
accessing and managing these resources.
5.2 Core Cloud Middleware
This layer offers the services like storage access, security, information registration and remote process management.
5.3 User Level Cloud Middleware
Also named as Cloud tools, this layer involves programming tools, scheduling the task, managing the resources and
developing the application environment.
5.4 Application Layer
This is the highest layer of the architecture. Since it helps the cloud users to see and interact, it is also called as service
ware.

International Journal of Application or Innovation in Engineering& Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 36


Figure 2 Cloud Architecture
6. SURVEY
The survey of DoS attacks in cloud says that as the use of cloud grows the rate of DDoS [7] attacks will also catch a fast
pace. In 2012, the eighth annual Worldwide Infrastructure Security Report, after examining a twelve-month period
resulted with the key finding that 14% had to deal with the DoS attacks targeting the cloud services. Last year, this
report also published that not only the companies but also the cloud providers are being infiltrated by these attacks. In
year 2008 only, these attacks were commonplace i.e. it became a game to keep them which had to be exposed on public
networks[8]. The survey of past few years says that the automated tools and the defense mechanisms built up for the
cloud services in order to prevent them from attacks were not that very useful. In November 16, 2012, 65 persons of
organizations experienced the DoS[9] attacks in the past twelve months. Survey also depicts that the companies always
live under the consecutive threat of a salt resulting in a huge damage. It came into notice that the infection by the
attacks is done within average downtime of 54 minutes per attack resulting in the cost revenue of 3.5 million $ per
year. The survey up till now shows that 60% ask to halt an attackers computer in the company, 75% rely on
antiviruses, 34% feels the lack of system visibility, and 31% are the negligent one. The respondents of the survey depict
that in a cloud computing model, highly affected sectors are public sector, financial department, pharmaceutical as well
as health care in the past 11 years[3],[10].
7. THE CLOUD SECURITY
As cloud is in commercial use now, so maintaining the users and resource security, integrity, confidentiality and
privacy is a prime issue. Cloud security[6] itself ensures many security challenges and defense mechanisms. The main
goal of the cloud security is to provide an easy access to the legitimate users by preventing them from the serious
attacks like DDoS[7] attacks which are degrading the performance of the cloud computing model.
The cloud intrusions are of four types. These are listed below:
1. Unauthorized Access
2. Misuse
3. Cloud Exploit
4. Host or Network Specific Attacks
In order to gain defense against these above listed cloud intrusions, this paper has proposed a cloud intrusion detection
system. This defense mechanism will first analyze the attack traffic features. The analysis can be extracted from the
packet header since it has a field called traffic feature. This defense mechanism then finds a metric that can extract the
attack traffic feature to be used in the detection system [8].
Some different types of DDoS attacks along with their effects and defense mechanisms occurring in the cloud
computing model are listed in the table given below:
Table 1: Cloud DDoS Attacks, their effects and solutions
S.No. Name of the attack Function of the attack Solution against the attack
International Journal of Application or Innovation in Engineering& Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 37

1. Bandwidth Attack

Consumes targets resources

Multops, tree of nodes, detects the disproportional
packets going and coming from the attacker

2. ICMP (Ping) Flood


Bandwidth attack that uses ICMP
packets
ScreenOS, providing a Screening option which
sets a threshold that once exceeded invokes the
ICMP flood attacks

3. Ping of death

Sends multiple malformed or malicious
pings to a computer

Add checks for each incoming IP fragment telling
whether the packet is invalid or valid.
4. Amplification attack

Attacker makes a request that generates
a larger response

Using high performance OS, load balancer,
limiting the connection, limiting the connection
rate.

5. DNS Flood

Attacks both infrastructure and DNS
application

Radware carrier solution, allowing continuous
DNS service even under the attack and mitigating
the DNS attack.

6. HTTP GET Flood

Attackers send a huge flood of requests
to the server and consume its resources

NS FOCUS provides web application firewall,
Intrusion prevention system, carrier-grade anti-
DDOS system.

7. Reflector Attack

Where third parties bounce the attack
traffic from attacker to the target

DERM (Deterministic Edge Router Marking),
helps in identifying, tracking and filtering the
attack.

8. Smurf Attack

Attackers use ICMP echo request
packet to generate DOS attacks

Ingress filtering, configuring all the hosts and
routers not to respond to ICMP requests and not to
forward the packets directly to broadcast
addresses.
8. THE CLOUD INTRUSION DETECTION SYSTEM
This defense mechanism includes two methods of intrusion detection: the Behavior-based method which compares the
recent user actions to the usual behavior and the Knowledge- based method which detects known attacks. The behavior
deviation is analyzed using artificial intelligence. With all the responses, the IDS detect the attack and alert the other
nodes.
The Cloud Intrusion Detection Model [11],[12] will be detecting the attack traffic with the help of Entropy and The
Anomaly based detection system.

8.1 Entropy
International Journal of Application or Innovation in Engineering& Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 38

Entropy, being an important concept of information theory will be measuring the randomness of the data coming over
the network. The value of entropy lies in the range [0, log n]. More random is the data, the more entropy it has. When
the data belongs to one class, i.e. the class distribution is less even, the entropy will be smaller. On the other hand, if
the class distribution is more even the entropy will be larger. In order, to detect the deviation in the behavior of the
packets, a threshold value of the entropy will be set. On comparing, the actual value of the entropy of the packet with
the threshold value, if the value of the entropy of the packet mismatches the threshold value, it will be resulting in an
attack or detecting the changes in the randomness [13].

8.2 The Anomaly-based Detection System
In the Anomaly-based Detection System, the change or deviation in the entropy is used for the detection of the attack
traffic over the IP addresses or ports. Here, routers will communicate with each other in order to detect the attack. The
Anomaly-based Detection System will be deployed in every router. The entropy of each router is calculated and
compared with the carefully set threshold value. The entropy rates of different routers in the network are same. On
being mismatched, the suspected DDoS attack from the attack traffic is identified. The Anomaly-based Detection
System has an advantage over the Signature-based Detection System because in ADS, a new attack can be detected
even if it falls out of the normal traffic behavior. With the help of above mentioned two measures, the attack can be
detected [14]. The flow chart for this defense mechanism is as follows shown in figure 3.

Figure 3 Flow Chart for attack detection in Cloud Intrusion Detection System
9. CONCLUSION
This paper takes into account the various aspects of cloud. Out of all these aspects, DDoS is the most challenging threat
to the cloud services. So, in order to manage and increase the security of cloud, we applied a cloud intrusion detection
system known as Entropy-based Anomaly Detection System. This defense mechanism helps the user to detect and block
the attack before reaching the victim with the high detection rate. If we carefully and perfectly set the threshold of the
entropy, then are proposed based ADS can competently detect the DDoS attacks.

International Journal of Application or Innovation in Engineering& Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com
Volume 2, Issue 3, March 2013 ISSN 2319 - 4847

Volume 2, Issue 3, March 2013 Page 39

REFERENCES
[1] I. Foster, and C. Kesselman, The Grid: Blueprint for a new computing Infrastructure, Morgan Kaufmann
publishers, San Francisco, USA, 1999.
[2] M. Zakarya, and A. Ali Khan, Cloud QoS, High Availability & Service Security Issues with Solutions, IJCSNS
International Journal of Computer Science and Network Security, Vol.12, July 2012, PP 71 72.
[3] J. Joseph, M. Ernest, and C. Fellenstein, Evolution of grid computing architecture and grid adoption models,
IBM systems journal, Vol. 43, No. 4, 2004.
[4] S. Kar, and B.Sahoo, An Anomaly Detection System for DDOS Attack in Grid Computing, International
Journal of Computer Applications in Engineering Technology and Sciences (IJ -CA-ETS), Vol. 1, Issue 2, April
2009, PP 553-557.
[5] S. Kar, and B.Sahoo, Security Issues and Preventive Model for Grid computing, National Conference on
Modern Trends of Operating Systems, (MTOS- 09), March, 2009, PP 35-38.
[6] I.Gul, and M. Hussain, Distributed Cloud IntrusionDetection Model, International Journal of AdvancedScience
and Technology, Vol. 34, September, 2011, PP 71-82.
[7] C. Douligeris and A. Mitrokotsa, DDoS attacks and defense mechanisms: Classification and state-of-the-art,
Computer Networks: the Int. J. Computer and Telecommunications Networking, Vol. 44, No. 5, April 2004, pp.
643666.
[8] G.Bhatti, R.Singh and P.Singh, A look back at Issues in the layers of TCP/IP Model, International Journal of
Enhanced Research in Management & Computer Applications, Vol. 1, Issue 2, November 2012.
[9] J. Mirkovic, S. Dietrich, D. Dittrich, and P.Reiher, Internet Denial of Service: Attack and Defense Mechanisms,
Prentice Hall PTR, December 2004.
[10] S. Shirasuna, A. Slominski, and D.Gannon, Performance comparison of security mechanisms for grid services,
Proceedings in 5th IEEE/ACM International Workshop on Grid Computing, 2004, PP 360-364.
[11] K. Vieira, A. Schulter, C. B. Westphall, and C. M.Westphall, IntrusionDetection for Grid and Cloud Computing
, IEEEComputer Society, Vol. 12, July/August 2010, PP 38-43.
[12] S. Roschke, F. Cheng, and C.Meinel,Intrusion Detection in Cloud, 8
th
IEEE International Conference on
Dependable, Automatic and Secure Computing, PP 729 734.
[13] G.Meera Gandhi, and S.K. Srivatsa, An Entropy Architecture for defending Distributed denial-of-service
attacks, International Journal of Computer Science and Information Security, Vol. 6, 2009, PP 129 136.
[14] G. Nychis, V. Sekar, D. G. Andersen, H. Kim, and H. Zhang, An Empirical Evaluation of Entropy based Traffic
Anomaly Detection, 10
th
International IFIPTC 6 Networking conference, 2011, PP 1 14.