International Journal of Computer Science Trends and Technology (IJCST) Volume 4 Issue 3, May - Jun 2016
RESEARCH ARTICLE
OPEN ACCESS
Self-Organizing Trust Model
Suyog Ashokrao Nagare [1], Prof. D.B. Kshirsagar [2]
Department of Computer Science and Engineering
SRES, College of Engineering, Kopargaon
India
ABSTRACT
As compared to other system Peer to peer system has open access, in networking domain. Each peer is capable of sharing
informat ion to other peer in peer to peer system. So there are chances of malicious activities increases. Along with
recommendations one peer must send trust parameters to another peer for better security. In this system, reco mmendations are
derived based on priority, trustworthiness, history, and peer satisfaction. The peer will co mmunicate with that peer only who is
having more recommendations and trustworthiness values.
Keywords :- Distributed systems, trust management, reputation, and security.
I.
INTRODUCTION
Peer to peer netwo rk is the collection of independent peers.
These peers share data among them without using any
centralize system. So security is the main issue. On this
system any malicious user can attack easily. To avoid the
malicious attack in peers we are maintaining trustworthiness.
But the challenging task is to keeping trust on another peer.
Because the opponent peer can be malicious. To state into the
numerical format, the peer is very co mplex as the trust is
logical and social phenomenon. For the file sharing between
peers, classification of peer as trustworthiness or non trustworthiness is not always efficient. So we are maintain ing
the matrices here for peer trust calculation.
The trustworthiness alone is not a sufficient approach to
communicate with peer. Thats why; along with
trustworthiness of the peer we maintain the reco mmendation
matrix and reputation. Here self-organizing trust model (sort)
technique focuses by maintaining trust relations among peers
in their surroundings to reduce malicious activity in a peer to
peer distributed system. In th is system fro m remain ing all
peers, it do not try to collect trust informat ion. Here about the
peers interacted in the past, every peer develops its own local
computation of trust[9]. Like this, good peers form dynamic
trust groups evaluated in their surroundings and from system
it can remove malicious peers.
The three matrices are calculate here. The reputation metric
is the first matric and based on peers recommendations this
matrix is calculated. Among all peers it is important while
deciding strangers and new nodes. Second, the primary
metrics to compute trust relation in the service and
recommendation surroundings are service trust metrics and
recommendation trust metrics. The service trust metric is used
while deciding service providers. Recommendation trust
ISSN: 2347-8578
metric is used while requesting recommendations. While we
evaluate the reputation metric, trust met ric reco mmendations
are calculated on the basis of recommendation
II.
EXISTING SYSTEM
To store and manage trust informat ion, in the presence of
an authority a central server is a p referred way. The trust
informat ion is securely stored by central server and defines
trust metrics. According to the trust informat ion the trust of
peer is detected from the central server.
In peer to peer system there is central manager, so that the
chances of malicious activities are increases. To imp rove
efficiency and the accuracy in the distributed system there are
many researches have been done already. Here by maintain ing
trust value and recommendation, we have propose new system
which mitigates risk management.
By malicious behaviours of peer entities the possible utility
loss is known as risk are caused by the potential security
policy vio lations. We generate trust relationships between
peers to guide security parameters with risk management[5]. It
helps to keep security at fixed level to the organization..
A. Algorithm for Reputation Management
Here some background protocols are exp lained. We only
explain the abstract model of so me network level model
protocol due to space limitations the detailed overview of each
protocol.
In this paper, all techniques are represented in terms of
modelled structure. On the evolutional parameters used in the
system our system performance and integrated security is
depend.
www.ijcstjournal.org
Page 380
International Journal of Computer Science Trends and Technology (IJCST) Volume 4 Issue 3, May - Jun 2016
B. Overlay routing model
To describe the overlays such as CAN, Chord, Tapestry and
Pastry the structured Distributed routing overlay have used
here. Here, fro m a large id space of integer data-type, uniform
node IDs are assigned by participating nodes. With unique
keys some objects are assigned.
Every key is assigned by the overlay to a unique live node,
which is called the keys root. Protocol routes packet is in
between this root and other peer. The routing table is
maintained the unique id of every protocol along with port
number. For sending packet in between nodes effectively this
routing table is very important. As well as a neighbor list is
maintained by each node wh ich consist of the unique id of
each node and number of nodes present at that particular peer
space.
C. DMRep
It is the system which is structural distributed system. Here
all nodes maintain their trust values by sending trust value to
the central authority. Two trust models which are defined
aberer and despotovics trust model, where by using p -grid,
peers report their co mplaints. The reputation based problem is
addressed here in data access layer. The peer is considered as
the trustworthy peer, until there is any co mp laint rises about a
peer. But this causes problem while adding new co mer inside
the existing network.
The main advantage of the peer to peer system is every peer
is capable of storing and sharing data to other nodes. This lead
uses of network bandwidth and reduction in storage costs. So
as others it considered as scalable.
The distribution of peer feedbacks is Power Trust [5]. On
the basis of their power nodes are ranked. Strongest node are
low power node. Fro m all adjacent peer, system calculates
global feedback to wh ich data sharing is dine. In this
community context is improved using factors which are
utilizing power nodes, feedback aggregation speed, and global
reputation accuracy. Its having advantages like, power law
distribution, fast reputation generation, ranking, system
robustness and efficiency, disadvantages are Non deployment
of power trust on unstructured nodes, fail to detect intrusions,
collusions, and selfishness of peers and failed to calculate
global trust value of each peer in a network.
To prevent malicious activity in the network, B. Bhargava
A burak et al. [7] gives a self-organizing trust model (sort).
No global informat ion is used. Trust informat ion of all peer
adjacent in netwo rk does not collect peer. Two matrices of
trust describes SORT, service and reco mmendation matrices
are defined on the basis of services provided by peer and
feedback received fro m peer. Feedback is considered as
recommendation which consist of level of reco mmendation
given by own peer. Local t rust informat ion is considered. In
this paper. Reputation queries [1] send by peers only to peers
interacted in the past. Disadvantages is that the system cannot
detect its trust value if any peer is starts to become malicious
after some time span.
III.
PROPOSED SYSTEM
PEER to PEER algorith ms enable a peer to reason about
trustworthiness of other peers based on past interactions and
recommendations. By using local informat ion available, Peers
create their own trust network in their pro ximity and do not
try to learn global trust informat ion. In proposed system peers
do not collect information of all pairs in the network they only
keep information of neighbours.
This system has following main roles:
A) Service trust matric
B) Reputation Trust Metric.
C) Recommendation Trust Metric.
A. Service Trust Matric:
Fig 1. P2P overlay network Architecture.[6]
D. Power Trust
ISSN: 2347-8578
Using the info rmation in its service h istory a peer first
calculates competence and integrity belief values when
evaluating an acquaintance's trustworthiness in the service
context. Ho w well an acquaintance satisfied the needs of past
interactions represented by Competence belief [10]? Let in the
service context the friend request denote the competence
www.ijcstjournal.org
Page 381
International Journal of Computer Science Trends and Technology (IJCST) Volume 4 Issue 3, May - Jun 2016
belief of Pi about
. Average behaviour in the past
interactions is a measure of the co mpetence belief.
Consistency is as important as competence. Integrity belief is
the level of confidence in predictability of future interactions.
IV.
Let in the service context, I
denote the integrity belief
of Pi about . A measure of the integrity belief is Deviat ion
from average behaviour (
).
1. First initialize the peers in the network.
B. Reputation Trust Matric:
ALGORITHM
The Recommendation algorithm is used here which is having
following steps.
2. Then initialize threshold value
3. Trust values Calculate threshold for recommendation.
4. Calculate the threshold from highly trusted acquaintances for
requests recommendations.
The reputation metric measures a stranger's trust
worthiness Based on recommendations. We assume that is
a stranger to Pi and
is an acquaintance of Pi In the
following two sections[4]. If Pi starts a reputation query to
collect reco mmendations from its acquaintances, if it wants to
calculate
value[2]. Trustworthy acquaintances and requests
their reco mmendations. Let the maximu m nu mber of
recommendations denoted by max that can be collected in a
reputation query and the size of a set S denoted by . Pi sets
a high threshold for reco mmendation trust values and requests
recommendations from highly trusted acquaintances first, in
the getting recommendation algorithm.
5. Evaluate Recommendation according to trust value of the
recommender.
6. Decreases the threshold and repeats the same process.
7. When maximum recommendations are collected, if excessive
network traffic then the algorithm stops.
V.
RESULTS
Fig 3. All peer feedback
Above figure shows the module describing the feedbacks of
all the peers.
Fig 2. System Architecture.
C. Recommendation Trust Metric:
Assume that a particular service want to get to Pi. Pj a
probable service provider and is a stranger to pi. Pi requests
recommendations to learn Pj's reputation, fro m its
acquaintances. Assume that reco mmendation send back to Pi
fro m Pk[8]. after collecting all reco mmendations Pi calculates
rij. Then, p k's reco mmendation evaluates Pi, and stores results
in RH ik, and also updates rtik. Assuming Pj is trustworthy
enough, Pi gets the service from Pj. Then, p i and stores the
results in SH ij, and updates stij by evaluating this interaction.
ISSN: 2347-8578
Fig 4. Neighborhood peer status
Above figure shows that, the one system is connected to
how many nodes. Here, the system s1 connected with only
itself.
www.ijcstjournal.org
Page 382
International Journal of Computer Science Trends and Technology (IJCST) Volume 4 Issue 3, May - Jun 2016
REFERENCES
[1] K. Aberer and Z. Despotovic, Managing Trust in a Peer2-Peer Information System, Proc. 10th Intl Conf.
Information and Knowledge Management (CIKM ),
2001.R. Caves, Multinational Enterprise and Economic
Analysis, Camb ridge Un iversity Press, Cambridge, 1982.
(book style)
[2] Reference 2 F. Cornelli, E. Damiani, S.D.C. d i Vimercat i,
S. Paraboschi, and P. Samarati, Choosing Reputable
Servents in a DISTRIBUTED Net work, Proc. 11th
World Wide Web onf. (WWW), 2002.
Fig 5. Peer Review
The above figure shows that the list of peers along with
peer port and peer IP.
VI.
CONCLUSION
For security decisions a new trust wh ich integrates
risk management. Based security enforcement the unique
feature of utility maximizat ion through risk management such
a new model offers. This is achieved by trust enhanced
security making process using both the current state of
knowledge on the trustworthiness of the entities and the risk
allocation for the given interaction. In doing so to guide the
security decisions, we enable the leveraging of the knowledge
on trust relationships such that while keep ing the security risk
at a defined level, the underlying application gains maximu m
utility. To the best of our knowledge, this paper is the first
which maximize the utility to integrate risk management for
trust based security decisions.
ACKNOWLEDGMENT
I would like to take this opportunity to express my
profound gratitude and deep regard to my Project Gu ide,Prof.
D. B. Kshirsagar (Head of Co mputer Engineering
Depart ment), for his exemp lary guidance, valuable feedback
and constant encouragement throughout the duration of the
project. I also want to thanks Prof. P.N. Kalwadekar(PG Coordinator) for his valuable suggestions were of immense help
throughout my project work. His perceptive criticis m kept me
working to make th is project in a much better way. Working
under both of them was an ext remely knowledgeable
experience for me.
ISSN: 2347-8578
[3] Reference 3 S. Kamvar, M . Sch losser, and H. GarciaMolina, The (Eigentrust) Algorithm for Reputation
Management in DISTRIBUTED Networks, Proc. 12th
World Wide Web Conf. (WWW), 2003.
[4] ] Reference 4 [4] L. Xiong and L. Liu, Peertrust:
Supporting Reputation-Based Trust for Distributed
Eco mmerce Co mmun ities, IEEE Trans. Knowledge and
Data Eng., vol. 16, no. 7, pp. 843-857, July 004
[5] B. Yu and M. Singh, A Social Mechanism of Reputation
management in Electronic Co mmunit ies, Proc.
Cooperative Information Agents (CIA), 2000
[6] Z. Despotovic and K. Aberer, Trust-Aware Delivery of
mposite Goods, Proc. First Intl Conf. Agents and
Distributed Computing, 2002
[7] F. Co rnelli, E. Damiani, S.C. Vimercat i, S. Paraboschi,
and P. Samarati, A reputation-based approach for
choosing reliab le resources in Distributed networks ,In
CCS02, Washington DC, USA 2002.
[8] K. Aberer, A. Datta, and M. Hauswirth, P -Grid:
Dynamics of Self- Organizat ion Processes in Structured
DISTRIBUTED Systems, Distributed Systems and
Applications, vol. 3845, 2005.
[9] R. Zhou and K. Hwang, Powert rust: A Robust and
Scalable Reputation System fo r Trusted Distributed
Co mputing, IEEE Trans. Parallel and distributed
Systems, vol. 18, no. 4, pp. 460-473, Apr. 2007.
[10] M . Gupta, P. Judge, and M. Ammar, A Reputation
System for Distributed Networks, Proc. 13th Intl
Workshop Network and Operating Systems Support for
Digital Audio and Video (NOSSDAV), 2003.
www.ijcstjournal.org
Page 383
