Download - 802.11 WLAN MAC Layer Modeling · TEAM 2: 802.11 WLAN MAC layer modeling Mentor: Radu V. Balan 802.11 WLAN MAC Layer Modeling Lisa Driskell Yejun Gong Xueying Hu Purdue University

TEAM 2: 802.11 WLAN MAC layer modeling

Mentor: Radu V. Balan

802.11 WLAN MAC Layer Modeling

Lisa Driskell Yejun Gong Xueying HuPurdue University Michigan Technological

UniversityUniversity of Michigan

Rashi Jain Mechie NkenglaNew Jersey Institute of

TechnologyUniversity of Illinois-Chicago

August 17, 2007


Outline

v Objectivev Overviewv ns-2v Results (.nam)v Results (.tr)v Conclusionsv Future Objectives


Project Objective: Analyze the relationships of the parameters for a modified EO Markov model and validate the model under certain assumptions with (ns2) network simulations.

Objective


Overview

MAC MAC

Packets Generated

IFQ

Agent/UDP Transmission

Agent/NullFinal

Destination

System Layout


p

Overview


Prob (packet dropped due to collision) =

Avg # of retries = )p1(p1

p L−−

1Lp +

pL(1-p) + pLp

…pb(1-p)…p2(1-p)p(1-p)1-pProbability

L…b…210Exact # retries

Previous Analysis of Markov Model

Overview


• Analyzed Markov model• Compared analytical results with computed

results to verify the analysis.• Use analytical results compared with network

simulations to determine whether the system can indeed be modeled with a Markov chain.

Outline

Overview


Ns-2

Ns-2 simulator

Ns-2 SimulatorInput

.tcl file

.tr file

Output Trace File

.nam file


Ns-2

Ns-2 simulator

• Closely relates to real world.

• A packet-based event-driven simulation.

• Can incorporate the wireless mechanism

• Allows for mobile stations

• Provides animations


Ns-2

Example of a ns2 simulation


Nam Trace

Nam Trace Format<type> -t <time> -s <source id> -d <destination id> -p <pkt-type> -e <extent> -c <conv> -a <packet attribute> -i <id> -k <trace level>

0~#pkts

i

AGT, MAC

k

cbr,ACK

0~NSTA

0~simTime

h, r, d, +, -

psttype


Nam Trace : Medium Busy Time

The packet is on the air from t, if at ttype=‘h’ AND k=‘MAC‘ (Hop)

The medium is busy in[t, t+pktTransTime]

The medium is busy in collision case.

TxFrame 1

TxFrame 2

BusyCollision


Nam Trace : Medium Busy Time

Medium Busy Time ? , if NSTA ? < critical numberMedium Busy Time ? , if NSTA ? > critical number


1-Hop# cket)Retries(pa# with

ThisNodepktsSendTo#

packet thisof Retries#(node)AvgRetries hisNodepktSendToT

=

=∑Numerical:

Theoretical:

But how to find p?

retries. #max theis L andy probabilitcollision theis p where

)p1(*p-1

p(node)AvgRetries L−=


1LcpktTransSuColpktDropDue

ColpktDropDuep +

+=

pktSend=

pktDropDueCol+

pktDropDueQueue+

pktDropDueEnd+

pktTransSuc

Type=‘h’ AND k=‘AGT’ AND p=‘cbr’

Type=‘d’ AND k=‘IFQ’ AND p=‘cbr’

LastHopTime+pktTransTime>sucTime

Type=‘r’ AND k=‘AGT’ AND p=‘ACK’


Results (.nam)

Average number of retries


Example of a .tr output


Results (.tr)

Average number of retriesFrom simulation

From calculation


Results (.tr)

Average number of retries for AP with varying packet period and offset


Average number of retries with varying packet period and offset

5 Stations


Results (.tr)

Comparison between of the average number of retries from calculation and simulation


Conclusions

Conclusions

• Estimated parameter ‘p’• Determined the saturation value• Established validity of the Markov model• Established importance of assumption of

synchronicity


Future Objectives

• Use the .nam and .tr files to extract information about– Average service time – Average wait time

• Use the above times to find relationships between parameters

• Compare computed results and simulated results for different packet arrival configurations

• Compute the throughput/delay

• Create a deterministic model

(Someone Else’s) Future Objectives


Questions

Questions?