0 / 12 ad hoc networking via named data michael meisel, lixia zhang ucla vasileios pappas ibm...

13
1 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim [email protected]

Upload: rafe-chandler

Post on 14-Jan-2016

216 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

1 / 12

Ad Hoc Networking via Named Data

Michael Meisel, Lixia Zhang UCLA

Vasileios Pappas IBM Research

ACM MobiArch 2010

Speaker : Conque, Kim

[email protected]

Page 2: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

2 / 12

Outline

• Introduction

• Existing solutions for ad hoc

• New direction of networking

• Listen First Broadcast Later (LFBL)

• Conclusion

Page 3: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

3 / 12

Introduction

• Design of current Internet protocol stack has a limit to adjust• Mandating that packet delivery is governed by the desti-

nation IP address

• Reasons • Mobile network is infrastructural-free

– Internet protocols are generally built with infrastructure support in mind

• Node mobility introduces a high degree of dynamics in node interconnectivity

Page 4: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

4 / 12

Existing solutions for ad hoc

• Existing routing protocols for ad hoc network • Proactive protocol: WRP, DSDV

• Reactive protocol: DSR, AODV

• Hybrid protocol: ZRP, HARP

• Sharing features of the protocols• All features are not suitable for wireless channel using

broadcast in nature

a. Using node assigned its own IP address

b. Routing based on the single best path to the given destina-tion IP

c. For crossing each hop, controlling the signal range

Page 5: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

5 / 12

Failings of the current ap-proaches

• Assigning IP addresses to moving nodes is difficult

• Hard to cover increasing number of mobile device through limited IP• Not available to fix the location and use aggregation

• Considering Inherent trade-off between the accu-rary of routing and overhead to keep consistent is needed

• Hard to decide which node will be the receiver

Page 6: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

6 / 12

Other solutions for ad hoc network

• Opportunistic routing • ExOR and MORE

• Delay Tolerant Networking (DTN) • Differences between DTN and NDN

– Using unique name that are used directly for delivery

– Automatically embracing ad-hoc networking and delay toler-ant networking without adding functional modules

Page 7: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

7 / 12

New direction for mobile network-ing

• Named Data Networking (NDN)• How to communicate using the 3-way exchange

1. Announcing the content name

2. Sending out ‘Interest’ packets

Contents

~/Alice

Contents

Page 8: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

8 / 12

Benefits of NDN

• Changing the communication semantics from “where” to “what”• Data name carried on the node do not necessary change

• Better than both landmark-based and geo-based routing solutions for ad hoc

• Facilitating the security development in the archi-tecture• End-to-end cryptographic signatures and encryption

helps to make the data security better

Page 9: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

9 / 12

NDN for Ad hoc networking

• Simplifying the implementation• Node can use application data name directly

• Interest packets can be forwarded along multiple paths towards potential data location

– Routing loop would not be caused, because PIT (Pending In-terest Table) keeps track of interest

• Addressing and caching fragments of application data• Available to accept the subsequent request for the same

file or a request for the retransmission

Page 10: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

10 / 12

Listen First, Broadcast Later (LFBL)

• New forwarding protocol for wireless ad hoc net-works• Using variation of NDN`s 3-way exchange

• The process of routing using LFBL

1. Name prefix announcement

2. Interest forwarding as a response

3. Data return

Intermidi-ate nodes

Listen First Broadcast Later

Page 11: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

11 / 12

Evaluation of LFBL

• To validate the performance of LFBL, comparing it with AODV on dynamic environment

• Using four different metrics: RTT, Overhead, De-livery rate, Total data transferred

• Result of the evaluation• LFBL delivers nearly 5times more data compared to

AODV

Page 12: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

12 / 12

Evaluation of LFBL

• Characteristic results for LFBL vs. AODV in detail

Page 13: 0 / 12 Ad Hoc Networking via Named Data Michael Meisel, Lixia Zhang UCLA Vasileios Pappas IBM Research ACM MobiArch 2010 Speaker : Conque, Kim cqkim@mmlab.snu.ac.kr

13 / 12

Conclusion

• Current routing solution for ad hoc networking has a pitfall in dealing with dynamic environment • Requiring the full or partial network topology for the

computation of best routes

• LFBL is a new forwarding protocol for the highly dynamic networks