A link layer scheme for reliable multicast in wireless networks l.jpg
This presentation is the property of its rightful owner.
Sponsored Links
1 / 42

A Link Layer Scheme for Reliable Multicast in Wireless Networks PowerPoint PPT Presentation


A Link Layer Scheme for Reliable Multicast in Wireless Networks. Thesis defense of: Aarthi Natarajan Advising Committee: Dr. Sandeep Gupta Dr. Partha DasguptaDr. Andrea Richa. Outline. Motivation Challenges Related Work: IEEE 802.11 Multicast, LBP, DBTMA System Model

Download Presentation

A Link Layer Scheme for Reliable Multicast in Wireless Networks

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


A link layer scheme for reliable multicast in wireless networks l.jpg

A Link Layer Scheme for Reliable Multicast in Wireless Networks

Thesis defense of:

Aarthi Natarajan

Advising Committee:

Dr. Sandeep Gupta

Dr. Partha DasguptaDr. Andrea Richa


Outline l.jpg

Outline

  • Motivation

  • Challenges

  • Related Work: IEEE 802.11 Multicast, LBP, DBTMA

  • System Model

  • Protocols: RDNP and M-RDNP

  • Simulation Environment

  • Performance Results: Wireless LANs, Ad hoc networks

  • Conclusions and Future Work

Mobile Computing and Networking Group

Arizona State University


Group applications l.jpg

Group Applications

Search and Rescue

Operation

Chat

Application

  • More Applications …

  • Military Operations

  • Emergency operations

  • Whiteboard Applications

NEED “RELIABLE” COMMUNICATION

Mobile Computing and Networking Group

Arizona State University


Why wireless l.jpg

Why Wireless ?

Motivation

Wireless Network : devices with wireless adapters communicating with each other using EM waves

  • Ease and Speed of deployment.

  • Wired network may not be possible.

    Wireless Network Architectures

Centralized or LAN

Distributed or Ad hoc

Wired network

Base Station

1.Collection of autonomous hosts

2. No Infrastructure

3. All hop wireless

1. All devices connect to base station

2. Infrastructure based

3. End hop wireless

Mobile Computing and Networking Group

Arizona State University


Problem statement l.jpg

Problem Statement

Motivation

  • To build a reliable link-layer protocol for multicast in single channel multi-access wireless networks

  • Reliability can be achieved at

    • End-to-end: across several hop.

    • Link level: across a single hop.

  • Why reliable multicast at the link layer[IG00]?

    • Allows local error recovery.

    • Improves throughput.

    • Conserves energy.

    • Reduces end-to-end delay.

link level

reliability

Destination

Source

End to end reliability

Mobile Computing and Networking Group

Arizona State University


Link level multicasting l.jpg

Link Level Multicasting

  • Repeated unicast transmissions

    • Redundant data, Wastes energy, Increases delay, Reduces throughput

  • Reliable Broadcast at the multicast address and filter at the receivers

    Design Issues:

  • Medium Access:

    • Wired Networks use CSMA/CD

    • Wireless Networks signal strength fades with distance

      • self interference, hidden terminals exposed terminals, capture effect

  • Error Recovery

  • Controlling the flow of feedback information

Sender

5 transmissions

Sender

1 transmissions

Mobile Computing and Networking Group

Arizona State University


Medium access issues l.jpg

Hidden Terminals

Nodes not within the senders range but within the receiver range

Causes collisions at the receivers

Collision detection cannot be used

Location dependant carrier sensing: Even if the receiver may experience collisions, the sender may not.

Self Interference: transmit signal flows into receive path

Capture Effect

Picks up stronger signal as long as the ratio of the stronger to weaker signal exceeds the capture threshold.

Signal from Node B

Capture

Threshold (SNRT)

>

Signal from Node G

Medium Access Issues

Challenges

Node B

Node O

Node G

HIDDEN TERMINAL

Can still pick up packet from Node B

Node O

Node B

Node G

Mobile Computing and Networking Group

Arizona State University


Error recovery and feedback control l.jpg

Error Recovery and Feedback Control

  • Local Error Recovery

    • High channel BER

      • Channel bit error rate can be as high as 1 in 104 or higher.

      • Almost 40% or more of the packets are in error when payload is 512b.

    • Retransmission based [TKP97]

      • ACK based : absence of ACK

      • NACK based : presence of NACK

      • Explicit retransmission requests : reception of retransmit request packet

    • FEC based

  • Controlling the flow of feedback from multiple receivers

    Battery Anemic

    • Size and weight limitation restrict the lifetime of the device battery.

    • Energy conservation techniques

Mobile Computing and Networking Group

Arizona State University


System model and assumptions l.jpg

Single channel multi-access networks

Single transceiver

Infrastructure-based as well as ad hoc

Packet loss : Bit errors and Collisions

Group membership maintained by the higher layer protocols

Two Ray Ground Propagation Model

Signal has to greater than the reception threshold to receive the packet correctly

The medium is perceived as busy as long as the signal is greater than the noise threshold.

System Model and Assumptions

Preliminaries

Mobile Computing and Networking Group

Arizona State University


Some related work l.jpg

Some Related Work…

Related Work

  • Solutions to Hidden Terminals

    • RTS-CTS based : Single Channel

      • Unicast: IEEE 802.11Unicast

      • Multicast: LBP, PBP, DBP

    • Busy Tone based : Two channels

      • Unicast: DBTMA [DJ98]

      • Multicast: IEEE 802.11MX [Sha02]

  • IEEE 802.11 Multicast

Mobile Computing and Networking Group

Arizona State University


Ieee 802 11 unicast com99 l.jpg

DIFS

SIFS

SIFS

SIFS

X

X

X

X

RTS

CTS

X

X

DATA

X

ACK

IEEE 802.11 Unicast [Com99]

Related Work

  • RTS-CTS

  • ACK based error recovery

  • Physical + virtual carrier sensing

  • DIFS, SIFS inter-frame space for prioritization of DATA

Sender

H

Hidden Terminal

Receiver

H

RTS

DATA

Sender

H

CTS

ACK

Receiver

H

Update NAV from CTS

Update NAV from RTS

Others

Mobile Computing and Networking Group

Arizona State University


Rts cts for multicast l.jpg

RTS-CTS for Multicast

Related Work

  • Several receivers : feedback collision

  • Try to eliminate the collision of feedback

  • LBP[KK01] – leader node sends the feedback information

  • DBP[KK01] – all nodes send out feedback after a certain random delay.

  • PBP[KK01] – every node sends out feedback with certain probability “p”.

  • BSMA[TG00b], BMW[TG00a], BMMM, LAMM [Shal02]

  • RTS-CTS does not solve all hidden terminal problems[XGB02]

RTS

H

CTS Collision

Mobile Computing and Networking Group

Arizona State University


Ieee 802 11 multicast com99 l.jpg

DIFS

IEEE 802.11 Multicast [Com99]

Related Work

  • Not Reliable

  • Hidden terminal problems

  • No local error recovery

DATA

Sender

Consume data

Group

Neighbors

Ignore data

Others

Mobile Computing and Networking Group

Arizona State University


Our protocol l.jpg

Our Protocol

  • Salient Features

    • Protocol RDNP

      • Deals only with local error recovery

      • No CTS packet.

      • Uses a NACK or collision of NACKs to prompt retransmissions.

      • NACKs do not contain any relevant information.

      • Does not suppress hidden terminals

    • Protocol M-RDNP

      • Mitigate the effect of hidden terminals

      • Reliable neighbors do not suffer from hidden terminals as long as sender is transmitting

      • Forces routing layer to build routes only using reliable neighbors

Mobile Computing and Networking Group

Arizona State University


Protocol rdnp l.jpg

DIFS

SIFS

SIFS

Protocol RDNP

Protocol

  • Good for wireless LANs when there are no hidden terminals

    • base station is the only node that can transmit multicast data.

  • Not so good for ad hoc networks because of hidden terminals.

RTS

DATA

Sender

NACK

Group

Neighbors

Without packet

Update NAV from RTS

Others

&

Group Neighbors

With packet

Mobile Computing and Networking Group

Arizona State University


Reliable and interference region l.jpg

CSI

CS

RX

RL

Reliable and Interference Region

Protocol

Reception range: Radius within which the signal is greater than the reception threshold

Noise Range: Radius within which the signal is greater than the noise threshold

Hey! I cannot transmit.

I am within A’s noise range

Hey! I can transmit.

I am not within A’s noise range

Reliable Neighbors: All neighbors within the collision free zone.

Unreliable Neighbors: All neighbors not in the reliable range

Node A

Node B1

Node B2

Node C1

Node C2

Booo Hooo!

I experience collisions

Yippee! I still receive A’s signal

Thanks to capture effect.

Mobile Computing and Networking Group

Arizona State University


Minimum reliable radius l.jpg

Minimum Reliable Radius

Protocol

Minimum RL ≈ 170m

when CS = 550m

  • Assumption: No two nodes “start” transmitting simultaneously.

  • Two simultaneous transmissions must be separated from each other by a distance of CS

  • Around a sender the maximum number of nodes which can be transmitting simultaneously is 6

Node E

Node F

CS

dER

dFR

CS

Node R

dAR

dDR

d

Node D

Φ

RL

CS

Node A

Node S

dCR

dBR

Node C

Node B

Mobile Computing and Networking Group

Arizona State University


Protocol m rdnp l.jpg

Protocol M-RDNP

Protocol

  • Force all routes to be formed using only reliable neighbors.

  • Thus transmissions use only reliable hops in which there are no hidden terminal problems.

  • Might use more number of hops to transmit to the same node

Mobile Computing and Networking Group

Arizona State University


An example l.jpg

An example

Protocol

RL ≈ 170m

Routes using IEEE 802.11 and RDNP at the MAC layer

Routes using M-RDNP at the MAC layer

1

1

2

2

3

3

Number of hops = 4

Number of hops = 6

Mobile Computing and Networking Group

Arizona State University


Simulation environment l.jpg

Simulation Environment

Results

  • Network Simulator [Net02]

  • Performance Metrics

    • Average Packet Drop Ratio per Node =

    • Average Energy Consumed per Node per packet =

  • Wireless LANs:

    • IEEE 802.11, LBP, DBP, PBP, RDNP

  • Ad hoc networks

    • Routing Layer: SPST [GBS00], SPST [Sri03] better than M-AODV, ODMRP, MST

    • IEEE 802.11, RDNP, M-RDNP

  • All simulation points averaged over 45 runs

    • Accuracy 5% confidence interval 99% [Jai91]

Number of packets dropped per node

Number of packets sent

Energy consumed per node

Number of packets recv

Mobile Computing and Networking Group

Arizona State University


Simulation results wireless lans l.jpg

2

2

2

4

3

1

1

4

1

With Unicast traffic

AVG DROP RATIO

NUMBER OF NODES

Simulation Results – Wireless LANs

Results

Stationary nodes

Mobile nodes

AVG DROP RATIO

AVG DROP RATIO

BER (X 10e5)

BER (X 10e5)

Stationary nodes with

explicit retransmission requests

END-TO-END DELAY

BER (X 10e5)

Mobile Computing and Networking Group

Arizona State University


Simulation stationary ad hoc networks l.jpg

2

2

3

3

2

2

1

1

1

1

Simulation - Stationary Ad Hoc networks

Results

Nodes = 10, Avg. neighbor density ≈4,3

Nodes = 20, Avg. neighbor density ≈6,4

AVG DROP RATIO

AVG DROP RATIO

BER (X 10e5)

BER (X 10e5)

Nodes = 30, Avg. neighbor density ≈8,5

Nodes = 40, Avg. neighbor density ≈10,6

AVG DROP RATIO

AVG DROP RATIO

3

3

BER (X 10e5)

BER (X 10e5)

Mobile Computing and Networking Group

Arizona State University


Simulation manets l.jpg

3

2

2

1

1

3

Simulation – MANETs

Results

Nodes = 30, Low BER

Nodes = 10, Low BER

AVG DROP RATIO

AVG DROP RATIO

Speed (m/s)

Speed (m/s)

Nodes = 10, High BER

Nodes = 30, High BER

AVG DROP RATIO

AVG DROP RATIO

Speed (m/s)

Speed (m/s)

Mobile Computing and Networking Group

Arizona State University


Simulation manets very high speed 100miles hr l.jpg

Simulation – MANETS (Very High Speed ≈ 100miles/hr)

Results

Nodes = 10, Speed = 80 miles/hr

Nodes = 30, Speed = 80 miles/hr

AVG DROP RATIO

AVG DROP RATIO

BER

BER

Nodes = 10, Speed = 150 miles/hr

Nodes = 30, Speed = 150 miles/hr

AVG DROP RATIO

AVG DROP RATIO

BER

BER

Mobile Computing and Networking Group

Arizona State University


Summarizing reliability l.jpg

Summarizing Reliability

  • Stationary Ad hoc networks

    • M-RDNP - “Good” for low neighbor density.

    • M-RDNP and RDNP – Statistically indifferent for high neighbor density, “better” than IEEE 802.11.

  • Mobile Ad hoc Networks Low/Moderate Speeds

    • M-RDNP – “Good” for low neighbor density.

    • IEEE 802.11 - “Good” for low BER and high neighbor density.

    • RDNP – “Good” for high BER and high neighbor density.

  • Mobile Ad hoc Networks Very High Speeds

    • All three statistically indifferent.

Mobile Computing and Networking Group

Arizona State University


Energy results l.jpg

Energy Results

  • The energy consumed for a retransmission is much higher than the energy consumed for a transmission.

  • For stationary ad hoc networks,

    • As the BER increases the energy consumed per packet is much higher for RDNP and M-RDNP owing to the increase in the number of retransmissions.

    • RDNP consumes more energy than M-RDNP because of high drop ratio hidden terminals.

  • For mobile ad hoc networks

    • As the mobility increases, the energy consumed also increases.

    • For low BER the energy consumed by RDNP and IEEE 802.11 is almost the same, because no energy is lost in retransmissions.

    • M-RDNP consumes the least energy for low BER because is does not lose packets due to hidden terminals.

    • For higher BER RDNP and M-RDNP consumes more energy because of retransmissions

Mobile Computing and Networking Group

Arizona State University


Conclusions l.jpg

Conclusions

  • RDNP and M-RDNP was proposed as a NACK based reliable multicast extension to IEEE 802.11

  • Reliable multicast is extremely desirable when channel BER is high.

  • Frequent changes in route caused by SPST, “not good” for the MAC layer.

  • Energy cost associated with retransmission very high.

  • For very high speed networks MAC layer is insignificant.

    Future Work

  • Addition of energy saving strategies

  • Adapt the MAC layer based on the network characteristics

  • Estimate the link metric for SPST based on the conclusions

Mobile Computing and Networking Group

Arizona State University


Thank you l.jpg

Thank You!

Questions ?

Mobile Computing and Networking Group

Arizona State University


References l.jpg

References

[Com 99] ANSI/IEEE Standard 802.11 Wireless LAN medium control (MAC) and physical layer (PHY) specifications, In 1999 Edition.

[DJ98] J. Deng, Z. J. Haas, “Dual Busy Tone Multiple Access (DBTMA): A New Medium Access Control for Packet Radio Networks”, In IEEE ICUPC’98, Italy, 1998.

[KK01] J. Kuri, S. Kasera, “Reliable Multicast in Multi-access Wireless LANs”, Wireless Networks, 7(4):359-369, July 2001.

[Net02] Network Simulator – ns-2, Available via http://www.isi.edu/nsnam/ns/, [Accessed on Aug 02]

[Sha02] Vikram Shankar, “A Medium Access Control Protocol with reliable multicast support for wireless networks”, Master’s Thesis, Arizona State University, Tempe, AZ 85287, December 2002

[SG03] Ganesh Sridharan and Sandeep K.S.Gupta, “Performance comparison study of self stabilizing routing protocols for mobile ad hoc networks”, In preparation

[GBS00] Sandeep K.S. Gupta, A. Bouadallah and P.K. Srimani, “Self Stabilizing Protocols for Shortest Path Tree for multi-cast routing in mobile networks”, In proceedings of LCNS:1900, Euro-Par’00 Parallel Proceedings, pages 600-604, 2000.

[TKP97] Fouad A. Tobagi and Leonard Kleinrock, “Comparison of Sender-Initiated and Receiver-Initiated Multicast Protocols”, In IEEE Journal on Selected Areas in Communication, April 1997.

[SHAL02] Min-Te Sun, Lifei Huang, Anish Arora and Ten-Hwang Lai, “Reliable MAC Layer Multicast in IEEE 802.11 wireless networks”, In Proceedings of International Conference on Parallel Processing, ICPP ’02, pages 527-536, August 2002.

[XGB02] K.Xu, M.Gerla and S.Bae, “How effective is the IEEE 802.11 RTS/CTS handshake in ad hoc networks”, In Proceedings of IEEE Globecom 2002.

[TG00a] Kent Tang and Mario Gerla. “MAC Layer Broadcast Support in 802.11 Wireless Networks”, In Proceedings of 21st Century Military Communication Conference, MILCOM’00, pages 544-548, 2000

[TG00b] Kent Tang and Mario Gerla. “Random Access MAC for Efficient Broadcast Support in Ad Hoc Networks”, In IEEE Wireless Communications and Networking Conference, WCNC 2000, pages 454-459, 2000

[TG01] Kent Tang and Mario Gerla. “MAC Reliable Broadcast Ad hoc Networks”, In Communications for Network Centric Operations: Creating the Information force. IEEE Military Communication Conference, MILCOM’01, pages 1008-1013, 2001

Mobile Computing and Networking Group

Arizona State University


Medium access issues30 l.jpg

Capture Effect

Picks up stronger signal as long as the ratio of the stronger to weaker signal exceeds the capture threshold.

Oops! I would like to transmit but cannot !!!

Signal from Node B

Capture

Threshold

>

Signal from Node G

Medium Access Issues

  • Exposed Terminals

    • Nodes within the senders range but not within the receivers range

    • Reduces throughput

Can still pick up packet from Node B

Node O

Node O

Node P

Node B

Node B

Node G

Node G

EXPOSED TERMINAL

Mobile Computing and Networking Group

Arizona State University


Why rts cts does not work l.jpg

Why RTS-CTS does not work ???

RX

dAB

Node A

Node B

Node C1

Node C2

Mobile Computing and Networking Group

Arizona State University


Leader based protocol l.jpg

DIFS

SIFS

SIFS

SIFS

RTS

ACK

DATA

CTS

Group

neighbor

Sender

Non group

neighbor

Group

Leader

Leader Based Protocol

Related Work

RTS

DATA

Sender

CTS

ACK

Leader

NCTS

NACK

Group

neighbors

Update NAV from CTS

Update NAV from RTS

  • Problems:

    • Leader Mobility reduces throughput

    • “Capture Effect” may hide NCTS and NAK from distant nodes

    • Incoming nodes may not have heard RTS/CTS exchange and may cause collision

    • Sender has to know the multicast group members a priori

Mobile Computing and Networking Group

Arizona State University


Busy tone solution to hidden terminals l.jpg

Busy Tone Solution to Hidden Terminals

Related Work

Cannot transmit because I sense a receiver busy tone

tone

RTS

Node O

Node B

Node G

Node P

  • Problems:

    • Extra hardware, more energy

Mobile Computing and Networking Group

Arizona State University


Area around a transmitter l.jpg

CSI

CS

RX

RL

Node A

Area around a Transmitter

Protocol

  • Collision Free Zone: The area around a transmitter in which receiver do not suffer from hidden terminals when the transmitter is transmitting data.

  • Collision Zone: The area around the transmitter within which receivers are within the range of the sender but might suffer from hidden terminals.

  • Interference Free Zone: The area around a transmitter within which no node transmits because of physical carrier sensing.

  • Interference Zone: The area around a transmitter within which nodes can cause hidden terminal problems for receivers in the collision zone.

Reliable Neighbors: All neighbors within the collision free zone.

Unreliable Neighbors: All neighbors not in the reliable range

Mobile Computing and Networking Group

Arizona State University


Calculate rl and csi l.jpg

Calculate RL and CSI

For CSI -

dAB

dBC

Node A

Node B

Node C

For RL -

Mobile Computing and Networking Group

Arizona State University


An example36 l.jpg

An example

Protocol

RL ≈ 170m

Routes using IEEE 802.11 and RDNP at the MAC layer

Routes using M-RDNP at the MAC layer

Number of hops = 4

Number of hops = 6

Mobile Computing and Networking Group

Arizona State University


Spst self stabilizing routing protocol l.jpg

SPST Self Stabilizing Routing Protocol

Related Work

  • Every node periodically sends out beacon messages

  • Using values in the beacon messages SPST builds routes to the root of the multicast group.

Mobile Computing and Networking Group

Arizona State University


Confidence interval l.jpg

Confidence Interval

  • Each sample mean is an estimate of the population mean

    • With k samples we have k estimates

    • Problem: get one from k.

    • Best is get probabilistic bounds

    • Two bounds c1 and c2 such that there is a high probability, 1-α, that the population means is in interval (c1,c2)

      Probability(c1≤μ≤c2) = 1-α

      (c1,c2) confidence interval

      α significance level(≈0)

      100(1- α) confidence level (≈100)

      1- α confidence coefficient(≈1)

Mobile Computing and Networking Group

Arizona State University


Energy stationary ad hoc networks l.jpg

Energy - Stationary Ad Hoc networks

Results

Nodes = 20, Avg. neighbor density ≈6,4

Nodes = 10, Avg. neighbor density ≈4,3

AVG Energy consumed per packet

AVG Energy consumed per packet

BER (X 10e5)

BER (X 10e5)

Nodes = 30, Avg. neighbor density ≈8,5

Nodes = 40, Avg. neighbor density ≈10,6

AVG Energy consumed per packet

AVG Energy consumed per packet

BER (X 10e5)

BER (X 10e5)

Mobile Computing and Networking Group

Arizona State University


Energy manets walking speeds l.jpg

Energy – MANETs (Walking Speeds)

Results

Nodes = 30, Low BER

Nodes = 10, Low BER

AVG Energy consumed per packet

AVG Energy consumed per packet

Speed (m/s)

Speed (m/s)

Nodes = 10, High BER

Nodes = 30, High BER

AVG Energy consumed per packet

AVG Energy consumed per packet

Speed (m/s)

Speed (m/s)

Mobile Computing and Networking Group

Arizona State University


Energy manets vehicular speeds l.jpg

Energy – MANETS (Vehicular Speeds)

Results

Nodes = 10, Low BER

Nodes = 30, Low BER

AVG Energy consumed per packet

AVG Energy consumed per packet

Speed (m/s)

Speed (m/s)

Nodes = 10, High BER

Nodes = 30, High BER

AVG Energy consumed per packet

AVG Energy consumed per packet

Speed (m/s)

Speed (m/s)

Mobile Computing and Networking Group

Arizona State University


Reliability manets very high speed 100miles hr l.jpg

Reliability – MANETS (Very High Speed ≈ 100miles/hr)

Results

Nodes = 10, Low BER

Nodes = 30, Low BER

AVG DROP RATIO

AVG DROP RATIO

Speed (m/s)

Speed (m/s)

Nodes = 10, High BER

Nodes = 30, High BER

AVG DROP RATIO

AVG DROP RATIO

Speed (m/s)

Speed (m/s)

Mobile Computing and Networking Group

Arizona State University


  • Login