You are on page 1of 19

Adhiyamaan College of Engineering, Hosur.

(Autonomous) Department of CSE

NEIGHBOR NODE DISCOVERY AND TRUST PREDICTION IN MANETs Phase 2 Review 3 Guided by, Dr.D.Thilagavathy,
Professor and Head, Department of CSE

Presented By, Suganya Devi.S


(AC11MCS017) II YEAR ME-CSE

OBJECTIVE
The objective of this project is
To discover the one-hop neighbors of every node present in the network. To compute the trust level of the discovered neighbors in order to identify the malicious nodes.

PROJECT OVERVIEW
Neighbor discovery using Directional Antenna

Network monitoring and Packet Analyzing

Trust computation using Neural Network training and classification

Packet Analyzing
Packet Analyzer
Monitors the nodes individually. Following information are collected
Number of packets generated Number of Packets received Packet Forwarding Ration Packet modification information.

Sample File
Total Generated Packets = 202 Total Received Packets = 201 Total forwarded Packets = 0 Total dropped Packets = 1 Packet Modified=0 Packet Delivery Ratio =0.99505

Trust based Security


Trust
Belief level that one node had about another node.
Trust values range from 0 to 1.

Importance of trust
To work cooperatively. To avoid selfish and malicious node. To exchange the confidential information with the neighbors.

Types of Trust Computation


Distributed Trust computation . Centralized Trust computation. Hybrid Method.

Proposed Trust System


Node monitoring Apply Weights to Recommendation

Neural Network based Trust Prediction

Trust Aggregation Module

Recommendati ons

Trust Record Update

Trust Propagation

Trust computing Algorithm


Proposed Trust Computing Algorithm
Distributed trust computation Hybrid trust computation
Uses both Direct and recommendation based trust.

Neural Network Based Approach for trust prediction. Weights are assigned based on the age of relationship and the neighbors trust value.

Trust Record
Each node maintains a trust record that contains the list of neighbors and their trust level.

Direct Trust computation


The direct observation technique
Every node is involved individually to predict the trust.

Methodology
performs network monitoring Collects the trace file generated by monitoring the network. Uses the Packet Delivery Ratio, No of Packets generated, No of packets forwarded information to calculate the trust value of a node directly.

Contd..
Neural Network Based Approach
Uses Back Propagation Algorithm. Trains the network with the data collected by monitoring the network. Uses Supervised learning algorithm. The parameters to train the BPN
Packet forwarding ratio No of Packets generated per unit time.

The network is trained until the error converges to 10-6 The training algorithm used is traingd. Once the network is trained it can be used to predict the trust level of the node.

Snapshot- NN Training

Training

Response graph- Training

Response graph - Testing

Recommendation based trust


Recommendations

Received from one hop neighbors only. Sends RReq message to receive the recommendations. Weights are assigned to the recommendations based on age of relationship and trust value. Weight matrix is used to assign the weights

Trust Aggregation
Recommendations from all the neighbors are aggregated by applying the weights. Trust computation formula TA (B)=QA (B)+ i=1 ,2,3..n Ri (B)

Weight Matrix and Trust Matrix


Age of relationship Very old Old Old New Old New Trust node level of Very high High Average Average Poor Poor the Weights given to the recommendation 1 0.8 0.6 0.5 0.2 0.0

Node ID 1 2 3

Trust Value 1 0.6 0.8

Age Relationship Very old New Old

of

Advantages
Advantages of Proposed trust model are
Overcomes the Selective misbehavior attack and bad mouthing attack. Avoids network congestion. Diminishes the influence of the malicious nodes by providing variable weights to the recommendation. Reduces the false recommendations or recommendation modification.

REFERENCES
Kannan Govindan, Member IEEE and Prasant Mohapatra, Fellow IEEE, Trust Computations and Trust Dynamics in Mobile Adhoc Networks: A Survey Pedro B. Velloso, Rafael P. Laufer, Daniel de O. Cunha, Otto Carlos M. B. Duarte, and Guy Pujolle, Trust Management in Mobile Ad Hoc Networks Using a Scalable Maturity-Based model, IEEE Transactions on Network and Service Management, vol. 7, no. 3, September 2010. Yan Sun, Wei Yu, Zhu Han, and K. J. Ray Liu, Trust Modeling and Evaluation in Ad Hoc Networks , Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20740. Asad Amir Pirzada and Chris McDonald, Establishing Trust In Pure Ad-hoc Networks , School of Computer Science & Software Engineering, The University of Western Australia Copyright (c) 2004, Australian Computer Society, Inc. This paper appeared at the 27th Australasian Computer Science Conference

PUBLICATIONS and CONFERENCE


Journal
Published a paper, titled Neighbor Discovery and Trust Prediction in MANETs in International Journal for Science, Engineering and Technology Research, Volume 2 Issue 1 on January 2013.

Conference
Presented a paper titled DISTRIBUTED HYBRID TRUST

MANAGEMENT USING BPN NEURAL NETWORKS IN MANETs in


International Conference on Intelligent Instrumentation, Optimization and Signal Processing organized by Karunya university on March 1 and 2.

Future Conference
Paper titled AN INTEGRATED DIRECTIONAL ANTENNA MODEL FOR NEIGHBOR DISCOVERY AND TRUST PREDICTION IN MANETs
at International Conference On Computing , communication And Advanced Network, On l5 - l7 MARCH 20l3 at Chennai Institute of Technology.

THANK YOU!!!

You might also like