The Incentive Secure Mechanism Based on Quality of Service in P2P Network

1
The Incentive Secure Mechanism Based on Quality
of Service in P2P Network

• Yuhua Liu, Yuling Li, Kaihua Xu, Naixue Xiong,
  Jong Hyuk Park, YS Lee

Department of Computer Science Huazhong Normal
University Wuhan, China Emailyhliu_at_mail.ccnu.edu.
cn 2009/11/29
2
Outline

• Introduction
• Research Statuses
• The characteristics of Incentive Secure Mechanism
  
• The Incentive Mechanism Based on QOS
• The Ideology of Domain Partition
• The Calculation of Trust Value
• Calculating Nodes Effective Contribution Value
• The Process of Selecting Service Provider
• The Analysis of Strategy
• Simulation Experiment
• The Impact of BE(fk) and GE(fk) on Trust Value
• The Success Rate of Transaction
• The Number of Free-riders
• Efficiency of transaction
• Summarize

3
Introduction

• P2P network breaks the traditional C/S mode, in
  this network, each node is equal logically, which
  is both a provider and recipient of resources.
  P2P file sharing business has exceeded the WWW,
  and becomes one of the most important internet
  application systems in the world.
• because of the nodes characteristics of
  anonymity and voluntary contribute resources in
  P2P networks, there exists two problems needed to
  be solved (1) free-rider phenomenon the
  majority of nodes lack enthusiasm for providing
  services and only enjoy resources and services of
  the system without contribution. (2) unreliable
  service This is shown from the existence of a
  large number of fraud applications as well as the
  unreliable quality of services of the network
  itself.

4
Research Statuses

• At present, there are three ways to inhibition
  free-riding behavior, such as incentive
  mechanisms, the game theory and the economic
  model. A lot of literatures only consider the
  number share files, the size of shared data, and
  the degree of data sharing and so on, but ignore
  the quality of shared resources in the network.
• Some studies take into account the service
  quality of nodes and design some trust models.
  The trust models existing based on P2P at present
  are PKI trust model, trust model based on local
  recommended, data signature and the overall trust
  model. The calculations of trust value by adding
  the evaluation value while omit the other
  important factors such as the size and the number
  of documents.

5
The characteristics of Incentive Secure Mechanism

• (1) making full use of the approaches of physical
  location in network to divide the network into
  several regions, which can reduce the overhead of
  network communication.
• (2) calculating the trust value of document in
  accordance with the evaluation of consumption
  nodes, as well as the quality, size and number of
  documents shared of the node.
• (3) setting two parameters the parameter ? is
  used to punish the bad evaluation caused by
  providing poor quality file, parameter ? can slow
  the speed of accumulating initial trust value.

6
The Incentive Mechanism Based on QOS (1/5)

• The Domain Partition
• In real P2P networks, most of the nodes often
  communicate more frequently with the partial
  node, and so in accordance with the approach of
  physical distance between nodes in P2P network,
  the nodes locating nearer are allocated the same
  domain.
• we select one or more nodes that have good
  physical performance (high bandwidth, high
  processing power, large storage capacity) as
  management nodes to store resource information.

7
The Incentive Mechanism Based on QOS (2/5)

• The Calculation of Trust Value
• stands for the evaluation
  when the consumption node vj downloads the file
  fk resource provided by the node vi

(1)
8
The Incentive Mechanism Based on QOS (3/5)

• Calculating Nodes Effective Contribution Value
• Contribution value of node is close related to
  the uploading file quality, size, and quantity,
  and the quality is described by the trust value
  T(fk).

(2)

• We can calculate attenuation value according to
  (3)

(3)
9
The Incentive Mechanism Based on QOS (4/5)

• Nodes effective contribution value D(vi, T) is
  related to UD(vi, T) and DD(vi, T). The
  normalizing constant a and ß respectively stands
  for rewarding factor and punishment factor, just
  as (4)

(4)
10
The Incentive Mechanism Based on QOS (5/5)

• The Process of Selecting Service Provider

(5)

• According to formula (5) we can see that e is
  associated with files maximum value and size. At
  certain moment, if the effective contribution
  value is bigger than e, downloading action can
  been carried on, or the node must wait for a
  moment to accumulate its effective contribution
  value and request for downloading later.

11
The analysis of Strategy

• The performance of Network adopt PING, PONG to
  help peer find and detect the presence or absence
  of other nodes. In addition, select several nodes
  owning good physical properties as a backup node.
  
• Collusion and slander formula (1) can promote
  the consumer nodes to make a fair evaluation to
  some extent and resist effectively the phenomenon
  of collusion and slander.
• Unreliable service and fraud behavior This
  paper adopts method of calculating trust value of
  nodes for types of resources, and select service
  provider according to trust value. The existence
  of the parameters ? will make trust value of
  nodes subject to tremendous loss.

12
Simulation Experiment (1/5)

• The Impact of BE(fk) and GE(fk) on Trust Value
• Fig 2 The trust values change when?3,?5
• Fig.3 The relationship between GE(fk) and trust
  value
• Fig.4 The relationship between BE and trust value

Fig.2
Fig.3
Fig.4
13
Simulation Experiment (2/5)

• In order to evaluate this new strategy, some
  simulations have been carried out. We assume that
  the network has 10,000 nodes, there are 100,000
  documents stored, and the length of the document
  is random. Our simulation is based on Gnutella
  protocol and Flooding searching algorithm.
• We use BA model to construct the topology and
  simulate the operation of P2P system on machine.
  Suppose that there are 70,000 sharing documents
  (70 of network files) and 3000 free-riders (30
  of network nodes) in the network at the beginning
  of experiment. It assumes that the earnings
  factor a is 0.1, and the attenuation factor ß is
  0.05.

14
Simulation Experiment (3/5)

• The Success Rate of Transaction

Fig.5 Compare the success rate of transactions in
three networks
15
Simulation Experiment (4/5)

• The Number of Free-riders

Fig.6 Compare the success rate of transactions in
three networks
16
Simulation Experiment (5/5)

• Efficiency of transaction

Fig.7 Average path length between random network
and our network
17
Conclusions

• This strategy can guarantee the quality of
  service as well as restrain free-rider
  phenomenon, so as to enhance the utilization of
  resources and the rate of successful transaction.
  
• But due to the level of research and time
  constraints, this issue still has many areas to
  improve and continue to explore. For example, If
  the file always gets good evaluation in the
  transaction, it will attract a lot of consumer
  nodes to download, which will result in the
  emergence of hot spots, which will make the
  network more vulnerable.
• In the future, further research should be done to
  deal with the free ride phenomenon and security
  issues in P2P network..

18
Reference

• Wang, Y, Vassileva J. Trust and Reputation Model
  in Peer-to-Peer Networks. Peer-to-Peer Computing,
  2003. (P2P 2003). C// Proceedings. Third
  International Conference on, 1-3 Sept. 2003
  150-157.
• Huu Tran, Hitchens M, Varadharajan V, Watters P.
  A Trust Based AccessControl Framework for P2P
  File-Sharing Systems. System Sciences, 2005.HICSS
  '05 C//Proceedings of the 38th Annual Hawaii
  International Conference on, 03-06 Jan. 2005
  302c-302c.
• Adar E , Huberman B. Free Riding on Gnutella.
  First MonDay, 2000, 5(10) 32-35.
• Li Guo, Zhenqi Yang. P2P Technology and Security
  Analysis. Network Security Technology and
  Application, 2005.5, 26-28.
• Emmanuelle Anceaume, Maria Gradinariu, Aina
  Ravoaja. Incentive for P2P Fair Resource Sharing.
  Proceedings of the IEEE International Conference
  on Peer-to-Peer Computing (P2P05) 253-260.
• Karakaya M, Korpeouglu I, Ulusoy O.A. Distributed
  and Measurement-based Framework Against Free
  Riding in Peer-to-Peer Networks// Proceedings of
  the 4th International Conference on Peer-to-Peer
  Computing. Zurich, 2004 276-277
• Ramaswamy L , Liu L. Free riding A New Challenge
  to Peer-to-Peer File Sharing Systems//
  Proceedings of the 36th Hawaii International
  Conference on System Sciences. Hawaii, 2003
  220-229.

19

• Gong Haigang, Liu Ming, Mao Yingchi etal.
  Research Advances in Key Technology of P2P-Based
  Media Streaming. Computer Research and
  Development, 2005, 42(12) 2033-2040.
• Rohit Gupta, Arun K. Somani. Game Theory As A
  Tool to Strategize As Well As Predict Nodes
  Behavior In Peer-to-Peer Networks. Proceedings of
  the 11th International Conference on Parallel and
  Distributed Systems (ICPADS'05) 244-249.
• Xavier P, Wentong Cai, Bu-Sung Lee. Employing
  Economics to Achieve Fairness in Usage Policing
  of Cooperatively Shared Computing Resources.
  Proceedings of 2005 IEEE International Symposium
  on Cluster Computing and the Grid(CCGrid 2005)
  326-333.
• Ma R T B , Lee C M , Lui J C S , Yau D K Y.
  Incentive P2P networks A Protocol to Encourage
  Information Sharing and Cont ribution. ACM
  SIGMETRICS Performance Evaluation Review, 2003,
  31 (2) 23-25
• J.Altman,PKI Security for JXTA Overlay Networks.
  Sun Microsystem,Palo Alto, Tech
  ReptTR-12-03-06,2003.
• F.Cornelli, Choosing Reputable Servents in a P2P
  Network. In Lassner.D ed. Eleventh International
  World Wide Web Conference,Hawaii,ACM
  Press,2002.7 441-449

20
Thank you!

• 2009-11-30