Wireless Network Simulator - Ptolemy Project

A Heterogeneous Approach for Wireless Network Simulations Jens Voigt Mobile Communications Systems Dresden University of Technology 01062 Dresden : (+49) 351 463 4246

: (+49) 351 463 7255

: [email protected]

1. Motivation: System Simulations of Mobile Cellular Networks 2. Wireless Network Simulator: Overview 3. Choosing a Model of Computation: Specifics of Mobile Cellular 4. Application: Simulation of a 60 GHz Indoor System Dresden U n i ve r s i t y o f Te c h n o l o g y

3rd Ptolemy Miniconference UC Berkeley, 2/19/99 J. Voigt

0

Mobile Communications Systems

System Simulations for Mobile Cellular Networks a network operator... has

limited

many users

must re-use the resources

number of resources causes

interferences

quality capacity

wants to offer needs

to balance the scales

buys

a Wireless Network Simulator (WiNeS)

Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems

1


Wireless Network Simulator (1)

Overview

simulator configuration system-module (GSM)

WiNeS-core-classes

(interface definitions) WinesBase isA() getClassName()

system-module (60 GHz indoor)

1

WinesBaseInStar

WinesBaseInChannel

NE* myParent_p

Channel* myParent_p

getParent()

getParent()

getParent()

NetworkElement ExtSendQ myReadQ

...

1

WinesBaseInNE

DEStar* myParent_p

ExtSendQ myWriteQ -unsign. int myNum.

Channel

animation

BasePropagation

BaseMovement

ExtSendQ myAnswerQ handleReads() handleWrites()

BaseResourceOccupat.

BaseAntenna

-Position myPosition getNumber()

system-module (UMTS)

getPosition()

ExternSource

setNumber()

RefireQ myRefireQ

handleAnswers()

ExtSendQ myNewNEQ

BaseNetworkAccess

handleRefire()

InternSource BaseModem

RefireQ myRefireQ IntSendQ myTriggerQ handleRefire()

simulation control (Ptolemy-DE)

simulator Dresden U n i ve r s i t y o f Te c h n o l o g y


2


Wireless Network Simulator (2) Object-Oriented Analysis 1...n network inherent 0...n activities 1...n

data stream generator

resource maintenance 1...n

1...n

1...n backbone network

user activities

1...n

1

1

1...n

0...n

1 ... n

1

1

network element with fixed access

modem

network access

0...1

1

1 1 activation

resource usage

instantiates 1

1...*

1...n

1

transmission channel 1

1...n

movement

1...n radio wave propagation

antenna

1 1...n resource occupation

1...n

1...n

1

Dresden U n i ve r s i t y o f Te c h n o l o g y

resource allocation

1

1

1


1...n

1

network element 1

1

environment

3

1


Model of Computation (1)

Multiple Layer Model system

channel

backbone network all network elements network element (prototype)

configuration

activation configuration

element .....

.....

.....

component

imperative

programming (WiNeS-classes)

other design environments (SDF, FSM,...)

other tools (Matlab™, SDT™,...)

Model of Computation can be different in each layer Dresden U n i ve r s i t y o f Te c h n o l o g y


4



Events on System Layer

backbone network

channel

many synchronous events

network elements activation

a few random events (e.g. poisson distributed)

NE Core

component 1 component 2

event-tags need not to be totally ordered through all layers Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems

5



Channel Access

parallel

ordered in time ➔ multiple access technique technique ➔ logical control channels (e.g. a measure cycle)

➔ one transmission channel ➔ many network elements

channel

cell 1 cell 2 cell 3

cyclic

sequential ➔ channel access as

➔ an event cycle occurs

sequential process


6


Model of Computation (4) synchronous timed system discrete event MoC

Alternatives communicating sequential processes

synchronous language CSP/π-/Fusion-calculus

➔ mutable system configuration possible

➔ fast compile time scheduling

➔ overspecifies model if applied in all layers ➔ run-time scheduling (overhead ≈10%) ➔ sequential DE has performance limits ➔ parallelization hard to implement

➔ number and order of events must not change

fits for random events works fine Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems


➔ automatic scheduling based on rendezvous ➔ mutable system configuration and concurrency inherent in the model

➔ mutable system configuration ➔ random activation

fits for channel access not tried yet 7

➔ non-determinism use event cycle to get determinism? not tried yet 3rd Ptolemy Miniconference UC Berkeley, 2/19/99 J. Voigt

Connection between Ptolemy and System-Modules Ptolemy

WiNeS-Core

stars as scheduler interface 1

System-Module implements methods

defines interfaces member

MoC dependent 1...n star

interface definitions in WiNeS-core are independent of MoC

1...n

calls methods in members 1 1 0...n porthole



8


Application: 60 GHz System Module 11 00 11 00

11 00 11 00

11 00 11 00

virtual cell

office

office

11 00 11 00 11 00

office

hallway

1111 0000 1111 0000 1111 0000

1111 0000 1111 0000 1111 0000

base station

111 000 000 111

111 000 000 111

virtual cell

office

00 11 00 11 11 00 00 11 00 11 00 11 00 11 00 11 00 11

virtual cell office

1111 0000 0000 1111

hallway 111 000

virtual cell

office

11 00 00 11 00 11 00 11 00 11

base station

hallway

research project: campus network @ 60 GHz

base station

✔ each room == one cell ✔ system simulations in Generic Office Environment ✔ radiowave prediction by incorporated on-line Radiowave Propagation Simulator (RPS) ✔ different activation and mobility models Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems

9


60 GHz System Module (1)

Layout

simulator configuration (VEM)

RPC

simulation (Ptolemy 0.x)

system layer Dresden U n i ve r s i t y o f Te c h n o l o g y


10


60 GHz System Module (2) Base-Stations:

HOFMap

System Configuration Mobile Stations:

DEDynMap

system configuration

element layer Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems

11


60 GHz System Module (3) On-line Animation and Trace traced mobile resource occupation interference simulation (C++-Ptolemy 0.x)

CORBA™

animation (Java-Swing

and PtPlot)

link quality radiowave rays mobile’s path Dresden U n i ve r s i t y o f Te c h n o l o g y

12



Summary and Future Work - heterogeneous approach in three senses: -

different mobile cellular systems are modeled in different system modules

-

use of a Multiple Layer Model allows for a different MoC in each model layer

-

-

-

Event-Cycle for channel access allows for different MoC on system layer and simplifies parallelization

ready to use: -

two system modules (GSM and 60 GHz indoor)

-

simulation control: Ptolemy 0.7.1 - Discrete-Event Domain

-

on-line animation written in Java™ and connected via CORBA™

-

simulator configuration via Ptolemy0.x-GUI

plans: -

simulation control which allows for concurrency (first choice: base on PtolemyII CSP / PI)

-

more generic GUI for animation/configuration (use of DIVA system visualization?)

-

new system modules: 3G-UMTS, 4G-research project (IBMS)

Dresden U n i ve r s i t y o f Te c h n o l o g y Mobile Communications Systems

13
