integrating forwarding and replication in dtn routing a social network perspective

Integrating Forwarding and Replication in DTN Routing A Social Network Perspective

  Vehicular Technology Conference (VTC Spring), 2011 IEEE 73rd

Li, Yong ; Cao, Yuan ; Li, Su ; Jin, Depeng ; Zeng, Lieguang

Outline

• Introduction• Motivation and overview• The IFR routing scheme

– Community detection and weight assignment– IFR algorithm

• Evaluation and result– Combrage trace– Reality trace

• Conclusions

Introduction

• In the past few years, several routing algorithms according to store-carry-forward mechanism have been proposed to improve the performance of DTN.

• There are opportunistic forwarding (or probabilistic forwarding) based approaches, which rely on probability metrics such as time elapsed since last encounter, social similarity, geometric distance and so on.

• These related routing algorithms can be classified to two types, one type is replicating messages and the other type is forwarding only a single copy of each message.

Introduction (cont.)

• In this paper, we deign a routing scheme, which Integrates Forwarding and Replication (IFR) in a simple form from the aspect of social networks.

Motivation and overview

• The existing DTN routing algorithms can be classified to two types. One is replicating packets and the other is forwarding only a single copy.

• Replication increases the message delivery ratio but at the same time increases the delivery costs, while forwarding maintains at most one copy of the message in the network but achieves low delivery radio due to the node mobility and uncertainty.

Motivation and overview

• There are two part of our solution, the first part is the community detection and nodes’ delivery weight assignment, and the second part is the routing algorithm which Integrates Forwarding and Replication.

Motivation and overview

• In the first part, the mobile modes are divided to different groups according to the social relations reflected by the contacts in a distributed way, and for each node and community pair, a delivery weight will be assigned with a value.

• Then, in the IFR, when a mobile node sends messages to nodes belonging to the same community, the replication mechanism is used. Otherwise, the forwarding mechanism is used based on the delivery weight.

Community detection and weight assignment

• There are some community detection algorithms such as SIMPLE and K-CLIQUE are proposed by Hui et al..

Community detection and weight assignment

• Admission criteria for SIMPLE.

Community detection and weight assignment

• Admission criteria for k-CLIQUE.

Community detection and weight assignment

• By averaging the all history intermeeting time of a node with any nodes in a community, we can obtain the average contact rate of the node with that community.

IFR algorithm

IFR algorithm

IFR algorithm

A

B

A

B

IFR algorithm

A

B

IFR algorithm

A

B

IFR algorithm

A

B

A

B

A

B

IFR algorithm

AB

IFR algorithm

AB

Evaluation and result

• Combrage trace.

Evaluation and result

• Reality trace.

Conclusions

• In this paper, we focus on designing a simple but efficient mechanism to integrate replication and forwarding in the DTN routing from the social networks perspective.

• By simulations of two-community case and multi-community case based on traces of Cambridge and Reality respectively, we show that our scheme although increases the message delivery cost, it has much more increasing of the message delivery success ratio.

每週一句

• The existing DTN routing algorithms can be classified to two types. One is replicating packets and the other is forwarding only a single copy.

• One, the other

• One, another, the other

• One, another, still another, the other