20
Elec 599 Report: Modeling Media Access in Embedded Two-Flow Topologies of Multi-hop Wireless Networks Jingpu Shi Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering Rice University May 03, 2005

Jingpu Shi Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Embed Size (px)

DESCRIPTION

Elec 599 Report: Modeling Media Access in Embedded Two-Flow Topologies of Multi - hop Wireless Networks. Jingpu Shi Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering Rice University May 03, 2005. Motivation. Fairness problems in Multi-hop wireless networks. - PowerPoint PPT Presentation

Citation preview

Page 1: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Elec 599 Report: Modeling Media Access in Embedded Two-Flow Topologies of Multi-hopWireless Networks

Jingpu Shi Advisor: Dr. Edward Knightly

Department of Electrical and Computer Engineering Rice University

May 03, 2005

Page 2: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Motivation Fairness problems in Multi-hop wireless networks.

Root cause: different and incomplete channel state information.

Those problems have not been very well understood. All Stations are in range

G. Bianchi. Performance analysis of the IEEE 802.11 distributed coordination function. IEEE Journal on Selected Areas in Communications, 18(3):535–547, March 2000.

Not all stations are in radio range ?

In this work, we view a network as a set of sub-graphs consisting two flows and characterize its media access.

Page 3: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Assumptions and Notations Identical transmission range and interference

range.

We only consider one-way flows.

A link is established if two stations are in radio range.

Aa is the first flow, Bb is the second flow.

Page 4: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

All Possible TopologiesA

ba

B

Aa

Ab

Ba

AB

Bb

A

ba

B

Aa

ab

Ab

Ba

Bb

A

ba

B

Aa

ab

AB

Bb

A

ba

B

AaBa

AB

Bb

A

ba

B

Aa

ab

Ab

BaBb

A

ba

B

Aa Bb

(1) (2) (3) (4)

(5) (6) (8)

(9) (10) (11) (12)

A

ba

B

Aa

ab

Ab

AB

Bb

A

ba

B

Aa

Ab

BaBb

A

ba

B

Aa

AB

Bb

(7)A

ba

B

Aa

ab

Bb

A

ba

B

AaBa

Bb

A

ba

B

Aa

ab

Bb

Ba

Page 5: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Scenario Classification Senders Connected (SC): scenarios 2-7, where

senders of each flow are in radio range.

Asymmetric Incomplete State (AIS), scenarios 11 and 12, where senders are disconnected, asymmetric connections between the two flows.

Symmetric Incomplete State (SIS), scenario 8, 9 and 10, where senders are disconnected, symmetric connections between the two flows.

Page 6: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Scenario LikelihoodAssumptions and Illustration

What’s the probability of each scenario occurring, giving the two flows are connected?

Spatial analysis, assuming the two flows are uniformly distributed in a region and border effect is negligible.

Equal distance.

Page 7: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Scenario LikelihoodResults for each scenario

Scenario 11 dominates when distance becomes large

Page 8: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Scenario LikelihoodResults for each group

AIS and SIS class are highly likely to occur when distance between two hops becomes large.

Page 9: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Hop distance distribution

Page 10: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Outline Motivation

Scenario identifications and their likelihood

Fairness simulations

Media access modeling

Page 11: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Performance Simulations With CSMA/CA protocol

Observations: SC-No fairness

problem. AIS-Both short-

term and long-term fairness problems.

SIS-Long-term fair, short-term unfair.

Root cause: different information about the channel.

Page 12: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Mobility and Fairness

Page 13: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Outline

Motivation Scenario identifications and their

likelihood Fairness simulations Modeling media access

Page 14: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Modeling Framework View at single station

Identify 4 different state idle channel channel occupied by successful transmissions channel occupied by a collision busy channel due to activity of other stations

Define probabilities Probability of the four stats and throughput of the station

Page 15: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Model AIS Class: Strategies and steps

Compute collision probability for the flow Aa.

Compute busy probability due to other transmissions for flow Bb

Use decoupling technique

Assume flow Bb never collides. Flow Aa never defers.

Page 16: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Model AIS ClassResults

With RTS/CTS Without RTS/CTS

Page 17: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Model SIS ClassSample topology and modeling strategy

We analyze short-term unfairness.

Main difficulty: the two transmitting nodes are tightly correlated.

A Markov chain model using bi-dimensional state description.

Page 18: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Model SIS Class: Strategies and steps

We represent the system state as pair (SA, SB), where SA and

SB denote the backoff stage of Sender A and B respectively. Transition probability of the Markov chain.

ri is the probability that a station transmits after one slot in backoff stage i. f is the duration of the first packets (RTS or DATA) transmitted.

After solving the Markov chain, we can compute the transition time from state (m, 0) to (0,m), where m is the maximum backoff stage.

Page 19: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Model SIS ClassResults (cont.)

(C1) RTS/CTS access, m = 6, CWmax = 1024.(C2) RTS/CTS access, m = 8, CWmax = infinity.(C3) Basic access, m = 3, CWmax = 1024.(C4) Basic access, m = 6, CWmax = 1024.

Page 20: Jingpu Shi  Advisor: Dr. Edward Knightly Department of Electrical and Computer Engineering

Thanks! Questions or Comments ?