nest pi meeting july 9-12, 2002copyright © vanderbilt university/isis 2002 prowler probabilistic...

10
NEST PI Meeting July 9-12, 2002 Copyright © Vanderbilt University/ISIS prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR Features: Event-driven Deterministic or Probabilistic Command line / GUI Visualization Statistics / Optimization Any number of motes / any application Fast application prototyping MATLAB-based

Upload: ezra-henderson

Post on 17-Dec-2015

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR

Features: Event-driven Deterministic or Probabilistic Command line / GUI Visualization Statistics / Optimization Any number of motes / any application Fast application prototyping MATLAB-based

Page 2: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

prowler can simulate:

distance

distance

‘ideal’

measured

Signal strength

Send_Packet Packet_Sent

Channel_Idle_Check

Waiting_Time Backoff_Time Transmission_Time

Radio Propagation Transmit, Receive, Collision Model for signal strength:

based on distance deterministic or probabilistic (fading)

location (LARGE VARIATION)

time (SMALL VARIATION)

asymmetric links random failures/external disturbances

TinyOS w/ MAC-layer Complete MAC-layer protocol Simplified set of

Events Commands

+ Application (plug-in)

Page 3: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

1. Topology Information for the Motes coordinates of the motes can change during simulation

(may be random)

2. Visualization definition (optional) eventsvisual effects

3. Application Code actions to events generate new events

case 'Packet_Received' if rand<0.7 Send_Packet(RadioStream(data, memory.signal_strength)); end

Application

% Event_name Target Color Size anim_def={... {'Init_Application',1,[0 0 0 ],small}, ... {'Packet_Sent', 2,[1 0 0 ],small}, ... {'Packet_Received', 3,[1 0 0 ],small}, ...

Topology = [1,1;1,2;2,1; 2,2];

mote_IDs = [1,2,3,4];

Page 4: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

Example Application: Flood

Commands

Simulator

switch event

case 'Init_Application'

signal_strength=100;

%%%%%%%%%%%%%%%%%%%% Memory initialized here %%%%%%%%%%%%%%%%%%%%%%%

memory=struct('send',1, 'signal_strength', signal_strength);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if ID==1 % first node starts flood

Set_Clock(1000)

end

case 'Packet_Sent'

% do nothing

case 'Packet_Received'

if memory.send

p=sim_params('get_app', 'P');

if rand<p

Send_Packet(radiostream(data, memory.signal_strength));

end

memory.send=0;

end

case 'Collided_Packet_Received'

% this is for debug purposes only

case 'Clock_Tick'

Send_Packet(RadioStream(data, memory.signal_strength));

end

MATLAB

Events

Page 5: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

prowler – Main GUI

Application andradio definition

Graphic Visualization

Event Monitor

transmit

wait

receive

Text messages

Memory Dump of mote ID# 2:

send: 0

signal_strength: 1

Page 6: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

Ideal transmissionfunction

prowler – Simulation Parameters

MAC-layerparameters

Fading effectparameters

Reception condition

Page 7: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

Goal: • Optimize performance

Start

Optimization with prowler

Example: Flood in 2D with randomized retransmission (10x10 grid)

Parameters:• s – signal strength • p – retransmission probability

Page 8: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

Optimization with prowlerSimple performance metrics:

Page 9: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

Optimization with prowlerComposed performance metrics (‘adjustable interfaces’)

Metrics are composed from the number of receiving nodes & consumed energy: SNKNKSNNE TRRECTRREC 2

21 100),,(

strength signal:

nodes ing transmitt#:

nodes receiving #:

S

N

N

TR

REC

Emphasis on accuracy Emphasis on power consumption

optimum

Page 10: NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002 prowler PROBABILISTIC WIRELESS NETWORK SIMULATOR  Features:  Event-driven

NEST PI Meeting July 9-12, 2002Copyright © Vanderbilt University/ISIS 2002

prowler - Conclusions Fast and easy prototyping in MATLAB Captures the event-driven nature of TinyOS Algorithm testing in

deterministic environment probabilistic environment dynamically changing environment

Easy debugging Application area:

communication protocols, routing arbitrary application prototyping optimization, parameter tuning

Provides visualization Easy to expand; plugins