Drahtloses Labor für das “Internet der Dinge”Wireless Laboratory for the “Internet of the...

1

Drahtloses Labor für das “Internet der Dinge” Wireless Laboratory for the “Internet of the Things”

Prof. Dr. Mesut Güne! Distributed, embedded Systems Insitute of Computer Science Freie Universität Berlin http://cst.mi.fu-berlin.de

52. DFN Betriebstagung Mobile IT Forum 03.03.2010

2

Overview

•  Motivation •  Internet of the Things •  Characteristics of Wireless Networks •  Study Environments for Wireless Networks •  DES–Testbed •  Experimentation •  What can be experimented? •  Performance Metrics of the DES-Testbed

Prof. Dr. Mesut Güne! " www.des-testbed.net " DES-Testbed

3

Motivation


4

Motivation

•  Vision: Communication in future •  Anytime •  Anyplace •  Anything "   The Internet of Things

•  Applications •  Emergency scenarios •  Community networks •  Smart traffic systems •  Environment monitoring •  Medical applications •  Entertainment •  …

•  Communication?


5

Motivation

•  Integrated, heterogeneous wireless networks •  Mobile ad-hoc network (MANET), Wireless mesh network (WMN), Wireless

sensor network (WSN), Wireless personal area network (WPAN) •  Mobile nodes ! dynamic network topology •  Dynamic infrastructure ! functions of the infrastructure ! high flexibility •  Nodes may fail ! adaptive approaches •  Automatic configuration ! no user intervention, self configuration •  Device heterogeneity (CPU, memory, energy, sensors, …) •  Multi-hop communication ! A path consists of multiple radio links

! Requirements to the software: distributed, adaptive, and robust

Internet

MANET Mesh backbone Sensor


6

Motivation

Study of wireless networks in real environments

•  Methodology? •  Later more info

Working Environment City Environment


7

Characteristics of Wireless Networks


8


•  Number of nodes: number of all devices in the network, like routers, gateways, or hosts. The larger the number of nodes in a network, the more difficult it is to manage the network.

•  Mobility: This key refers to mobile nodes in the network, for example mobile routers and mobile clients. A network with a higher degree of mobility usually exposes a higher dynamic topology.

•  Hop-Count: number of hops between a source and destination. A high hop-count is likely to increase the latency of transmissions and decrease the throughput of a network.

•  Self-Organization: degree of human interaction required by a network, e.g., for configuration and management. Thus a network with a higher degree of self-organization is a network which demands less human interaction.

•  Energy-Awareness: energy sensitivity of a network. A network has to be more energy-aware if the energy resource is finite.

•  Universality: Characterizes whether the network is tailored to a specific application. A network is more universal if it can be used for more applications.

•  Data rate: user-perceived throughput, for example the quality of a connection from a source to a destination. Usually, the higher the data rate, the better the connection throughput. However, this key has to be used carefully, since a wireless link may show low quality due to interference even with high data rates.


9



10

Motivation Characteristics of Wireless Networks


11



12

Study Environments for Wireless Networks


13

Motivation

•  Scientific approach (networking, particularly wireless) •  Think of a problem •  Develop a solution •  Performance evaluation

•  Performance evaluation is mainly based on experiments •  Experiments are pillars of scientific work

•  What kind of experiments?

•  Trend: From simulation to testbeds! •  US, EU, and Japan have recently started large testbed based

projects (GENI, FIRE, AKARI, G-Lab) •  Focus: How are experiments run on a testbed?


Theoretical analysis

Application

Emulation

Simulation

Testbed

Virtualization

14

Study Environments

•  Applicability: Evaluates the degree of transferability of the results, conclusions, and the study environment into the real world.

•  Repeatability: Rates how straightforward the repetition of a given experiment in that study environment is.

•  Controllability: Assesses the degree of control the researcher has over the study environment as well as the studied subject.

•  Maintainability: Describes the effort to maintain the evaluation environment.

•  Scenario creation: Describes the freedom in creating different experiment scenarios.

•  Scalability: Assesses the feasibility of large scale experiments with respect to the number of nodes in the network, the experiment duration, and the number of network connections during the experiment.

•  Duration: Describes the experiment time. •  Cost: Evaluates the cost of experiments.


15

Study Environments Characteristics


Characteristic

Environments Theoretical

Analysis Simulation Emulation Virtualization Testbed

Applicability ! low middle high high

Repeatability ! high low low low

Controllability ! high middle middle low

Maintainability ! low middle middle high

Scenario creation ! simple middle middle hard

Scalability ! high middle middle low

Duration ! variable realtime realtime realtime

Cost ! low middle middle high

16

Study Environments Support of Network Layers


Layer (ISO/OSI)

Environments Theoretical

Analysis Simulation Emulation Virtualization Testbed

Application ! low high high high

Transport ! low/high middle high high

Network ! low middle high high

Data Link ! high middle middle high

Physical ! high/low middle low high

17

DES–Distributed Embedded System DES-Testbed


18

DES-Testbed

•  Testbed of the CST Research Group •  work in progress •  for student and research work •  focus on real world aspects •  comparison of experiments with simulation results •  transformation of theoretical approaches into software and

real systems •  interaction and integration of networking technologies

•  Weakness of current approaches •  Testbed in the box •  Short experiment run times

•  Important aspect @ CST •  Long-term studies


19

DES-Testbed

•  Required for •  industrial adoption •  long-term deployments (e.g. facility monitoring) •  safety critical applications (e.g. medical applications) •  security critical applications (e.g. intrusion detection)

•  Current approach •  few information about long term experiments •  no reliable or incomplete information about experiments •  limited reproducibility •  longest published times about 3-6 months (continuous?) •  field deployments often small, rarely up to 150 (just

present or functioning?)


20

DES-Testbed Components

•  Components of the DES-Testbed •  DES-Node: Consisting of wireless mesh and sensor node •  DES-Mon: SNMP based network monitoring tool •  DES-Web: User interface for experimenters •  DES-Exp: Experiment manager responsible for the

scheduling and execution of experiments •  DES-Cript: Experiment description language •  DES-Eval: Automatically evaluation of experiments •  DES-Vis: Visualization of the network

•  Goal: Building of an experimentation facility comfortable as a simulator


21

DES-Testbed Components

•  Distributed Embedded Systems (DES)

•  Goal of the DES-Testbed •  >100 hybrid nodes •  IEEE 802.11 network cards •  ScatterWeb MSB-A2 nodes •  Customized routers •  Easily expandable and

upgradable •  Spanning at least 3 buildings •  Indoor nodes •  Outdoor nodes •  Mobile nodes •  Wired connection to a server •  Collocated to university WLAN


DES Node

WSN WMN

22

DES-Testbed Architecture


23

DES-Testbed Characteristics


24

DES-Testbed DES-Node

•  PC Engines Alix2c2: •  500 MHz AMD Geode LX800 •  256 MB DDR DRAM •  2 Ethernet Ports

•  Via VT6105M

•  2 miniPCI slots •  dual USB 2.0 port •  CompactFlash socket •  Customized enclosure


25

DES-Testbed DES-Node

•  ScatterWeb Sensor node: •  Modular Sensor Board

Architecture (MSB-A2) •  LPC-2387 ARM7

•  98 kB RAM •  512 kB Flash

•  Chipcon CC1100 •  10dBm •  ISM band at 868 to 870 MHz •  max. data rate of 500 kbps

•  Coulomb counter for battery depletion measurement

•  GPIO pins •  mini USB 2.0 port •  microSD-card socket


26

Experimentation


27

Experimentation State of the Art

•  Manual conduction of large experiment series is very labor intensive

•  Many pitfalls in design and run of experiments "   Need for an automatic experimentation framework

•  Consider criteria for sound experimentation

•  The method of experimentation •  Design •  Implementation •  Evaluation

•  Problem: Experiment invention and measurement!


28

Experimentation Criteria for Soundness

•  Reproducibility The reported results should be reproducible by other researchers. This requires a detailed description of the experiment setup, study environment, and results.

•  Repeatability The effort to repeat a particular experiment should be minimized.

•  Unbiased The results should reflect a general idea of the subject of the study and should not be specific to an experiment.

•  Rigorous The experiment setup must reflect the true character of the subject to study.

•  Statistically sound The analysis of the experiment results must be based on mathematical methods.


29

Requirements for Automated Experimentation

•  Requirements for automated experimentation •  Workflow for experimentation •  Experiment description and definition •  Automated scheduling and running •  Time based running of experiments •  No or minimum load on the testbed nodes •  Support of testbed nodes without Ethernet connection •  Web based user interface •  Evaluation and preparation of results

•  Approach: DES-TBMS •  DES-Testbed Management System


30

DES-TBMS Experiment Workflow


31

DES-TBMS Components


32

DES-Cript

•  Domain specific language for experiment definition and description •  XML based •  Used by DES-TBMS for

scheduling and running


<experiment> <general> <name>...</name> <start_time>...</start_time> <owner>...</owner> <description>...</description> <iterations>...</iterations> ... <groups> <group name="groupname" role="Server"> <members>...</members> </group> </groups> ... </general>

<init> <action id="0"> ... </action> ... </init>

<actions> <action_block id="1"> <action id="1"> <group>...</group> <command>...</command> <start_time>...</start_time> <duration>...</duration> <evaluation_script>...</evaluation_script> </action> ... </action_block> ...

</actions> </experiment>

33

DES-Cript


<experiment> <general> <name>Simple TCP flow with iperf</name> <start_time>2009-09-11 20:00:00</start_time> <iterations>30</iterations> ... <groups> <group name="Receiver-TCP" role="Server"> <members><node id="t9-150"></node></members> </group> <group name="Sender-TCP" role="Client"> <members><node id="t9-155"></node></members> </group> </groups> </general>

<actions> <action_block id="1" execution_mode="1"> <action id="1"> <group>Receiver-TCP</group> <command>iperf -s</command> <start_time>0</start_time> <duration>330</duration> <evaluation_script>iperf.py</evaluation_script> </action> </action_block> <action_block id="2" execution_mode="1"> <action id="1"> <group>Sender-TCP</group> <command>iperf -c t9-150-wlan0 -t300</command> <start_time>10</start_time> <duration>330</duration> <evaluation_script>iperf.py</evaluation_script> </action> </action_block> ... </actions>

</experiment>

Actions can be executed in three modes:

•  Sequential •  Parallel •  Time based

34

DES-Web

•  Web based user interface for DES-TBMS •  An experiment is defined over this interface •  Alternatively, experiment definition can be uploaded

•  Multi user support ! parallel experiment definition •  Up-/download of files

•  Applications •  Log files

•  Log of the experiment run

Screenshot of the DES-Web


35

DES-Web


36

DES-TBMS Other Components

•  DES-Exp •  Automatic experiment

execution •  Preparation (Configuration)

•  Distribution and initialization of testbed nodes

•  New or specific software

•  Scheduling and Execution •  Logfile gathering •  Clean up

•  DES-Mon •  Monitoring the DES-Nodes •  Based on SNMP

•  DES-Vis •  Java-based 3D-Visualization

tool •  Network state (connectivity) •  Playback based on

experiment trace

Screenshot of DES-Vis showing the OLSR routing tree


37

DES-Testbed DES-Vis


38

DES-Testbed DES-Vis

39

DES-Testbed DES-Vis (Animation)


Color = Channel, Line thickness = Quality of channel

40

What can be experimented?


41

DES-Framework


Experiment (part of a study)

Experiment definition

Testbed Simulator Virtualizer

42

DES-Framework


Testbed Simulator Virtualizer

Performance Metric

43

Network Configurations


Wireless Sensor Network (WSN)

WMN as Integrator

for WSN

Multiple Portals

Internet Interconn

ection

Wireless Mesh

Network (WMN)

44

What can be experimented?

•  In fact, everything that runs on a Unix/Linux machine •  Particularly

•  Routing (DES-SERT) •  AODV, OLSR, ARA, BATMAN, NAMTAB

•  Localization •  Channel allocation •  MAC •  Applications: Audio, Video, FTP, etc.

•  Special configurations for wireless networks •  Wireless mesh networks (WMN) •  Wireless sensor networks (WSN) •  Integration of WMN, WSN, and the Internet


45

Performance Metrics of the DES-Testbed


46

Availability of DES-Nodes


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

a6-0

08a6

-009

a6-0

15a6

-017

a6-0

31a6

-032

aa6

-032

ba6

-102

a6-1

07a6

-108

aa6

-108

ba6

-124

a6-1

26a6

-139

a6-2

07a6

-212

aa6

-213

a6-2

15t9

-004

at9

-004

bt9

-004

ct9

-006

t9-0

07t9

-009

t9-0

11t9

-018

t9-0

20t9

-022

at9

-035

t9-0

40t9

-105

t9-1

06t9

-108

t9-1

11t9

-113

t9-1

17t9

-124

t9-1

34t9

-136

t9-1

37t9

-146

t9-1

49t9

-150

t9-1

54t9

-155

t9-1

57t

t9-1

58t9

-160

t9-1

62t9

-163

t9-1

65t9

-166

t9-1

69t9

-k21

at9

-k21

bt9

-k23

t9-k

40t9

-k60

at9

-k60

bt9

-k61

t9-k

63

Availability of DES-Nodes (01.01.10 - 01.03.10)

Up TimeDown Time

Was stolen and given back later!

47

Wireless Connectivity Performance Metrics of the DES-Testbed


48

Node Degree All W-Interfaces


0

5

10

15

20

25

30t9

-k40

a6-1

24t9

-117

a6-1

26a6

-017

a6-0

15a6

-213

t9-0

06a6

-009

t9-1

11a6

-032

bt9

-106

a6-1

08a

t9-1

60a6

-215

a6-0

31a6

-139

a6-1

07t9

-007

t9-1

62a6

-008

t9-1

57t

a6-1

08b

t9-1

08t9

-009

t9-1

65t9

-k21

bt9

-113

t9-0

11t9

-163

t9-k

60a

t9-1

54t9

-146

t9-0

18a6

-212

at9

-169

t9-k

21a

t9-1

34t9

-040

t9-0

04a

t9-0

04c

t9-0

35t9

-158

t9-0

22a

t9-1

50t9

-k63

t9-1

66t9

-020

t9-k

23t9

-124

t9-1

55t9

-137

t9-1

36t9

-k60

bt9

-k61

t9-0

04b

t9-1

49

Nod

e de

gree

49

Node Degree Per W-Interface


50

Node Degree Histogram Per W-Interface


0

5

10

15

20

25

30

35

40

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Freq

uenc

y

Node degree

51

Node Degree With Respect to Link Metrics


0

5

10

15

20

25

30

35

t9-165t9-117a6-015t9-006a6-126a6-213a6-032bt9-111a6-017a6-124t9-106a6-009t9-169a6-107t9-k21bt9-105t9-162a6-108aa6-031a6-108ba6-008t9-108t9-160t9-163a6-215t9-007a6-139t9-k60at9-004at9-009a6-212at9-011t9-k63t9-018t9-157tt9-146t9-040t9-004ct9-113t9-166t9-154t9-150t9-k23t9-124t9-k21at9-k61t9-136t9-158t9-022at9-020t9-134t9-155t9-137t9-k60bt9-004bt9-035t9-149

Nod

e de

gree

BPINGETX

52

Node Degree With Respect to Link Metrics


0

0.5

1

1.5

2

2.5

3t9

-165

t9-1

17a6

-015

t9-0

06a6

-126

a6-2

13a6

-032

bt9

-111

a6-0

17a6

-124

t9-1

06a6

-009

t9-1

69a6

-107

t9-k

21b

t9-1

05t9

-162

a6-1

08a

a6-0

31a6

-108

ba6

-008

t9-1

08t9

-160

t9-1

63a6

-215

t9-0

07a6

-139

t9-k

60a

t9-0

04a

t9-0

09a6

-212

at9

-011

t9-k

63t9

-018

t9-1

57t

t9-1

46t9

-040

t9-0

04c

t9-1

13t9

-166

t9-1

54t9

-150

t9-k

23t9

-124

t9-k

21a

t9-k

61t9

-136

t9-1

58t9

-022

at9

-020

t9-1

34t9

-155

t9-1

37t9

-k60

bt9

-004

bt9

-035

t9-1

49

Nod

e de

gree

Ratio of ETX to BPING

53

Path Length From each node to all others


1

2

3

4

5

6

7

8

9

10

t9-149t9-155t9-166t9-154t9-004bt9-137t9-035t9-150t9-136a6-139t9-k60bt9-169t9-158t9-004ca6-212at9-146t9-108t9-k61t9-134t9-004at9-022at9-040t9-020t9-157tt9-124t9-k63t9-163t9-k23t9-k21at9-160t9-162t9-113t9-105a6-215t9-106a6-031t9-007t9-k60aa6-213t9-111t9-018a6-032bt9-011t9-009t9-006t9-k21ba6-124a6-017a6-126a6-015a6-009a6-008a6-108ba6-108aa6-107

Network at 2010-02Network at 2009-09

Extension of the DES-Testbed

54

Synchronization and Clock drift Performance Metrics of the DES-Testbed


55

Clock drifts Without NTP, 3h Experiment


-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0 20 40 60 80 100 120 140 160 180

Diff

eren

ce [s

econ

ds]

Duration [minutes]

Time drift without ntp

158g158l158j

158h158m158k158i

56

Clock drifts Without NTP, 8h Experiment


-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0 50 100 150 200 250 300 350 400 450 500

Diff

eren

ce [s

econ

ds]

Duration [minutes]


158g158l158j

158h158m158k158i

57

Clock drifts Without NTP, 66h Experiment (Weekend)


-1

-0.5

0

0.5

1

0 500 1000 1500 2000 2500 3000 3500 4000

Diff

eren

ce [s

econ

ds]

Duration [minutes]


158g158l158j

158h158m158k158i

58

Clock drifts With NTP, 3h Experiment


-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0 20 40 60 80 100 120 140 160 180

Diff

eren

ce [s

econ

ds]

Duration [minutes]

Time drift with ntp

158m158k158i

158g158l158j

158h

59



-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0 50 100 150 200 250 300 350 400 450 500

Diff

eren

ce [s

econ

ds]

Duration [minutes]

Time drift with ntp

158m158k158i

158g158l158j

158h

60



-0.004

-0.003

-0.002

-0.001

0

0.001

0.002

0.003

0.004

0 100 200 300 400 500 600

Diff

eren

ce [s

econ

ds]

Duration [minutes]

Time drift with ntp Same measurement as on the last slide, but higher zoom!

61

Summary

•  DES–Testbed •  Comfortable playground for wireless network research •  Sound scientific methodology

•  Automated experimentation approach •  Experiment description language

•  DES-Framework (Testbed, Virtualizer, Simulator)

•  Involved in projects •  EU FP7 OPNEX •  EU FP7 WISEBED •  BMBF G-Mesh-Lab •  BMBF Wi-Mesh-Lab


Drahtloses Labor für das “Internet der Dinge”Wireless Laboratory for the “Internet of the...

Documents

Transcript of Drahtloses Labor für das “Internet der Dinge”Wireless Laboratory for the “Internet of the...