Sie sind auf Seite 1von 32

NetMap

A Network Measurement System for Android


Using Crowd Sourcing
Lars Mller Mikkelsen (lmm@es.aau.dk)
Ph.D. Student on the MOBiNET project
Wireless Communication Networks

Concept
Measuring network performance by using crowd
sourcing
Measured parameters relate to regular network usage

Goals
Measurement based network performance map
Statistics about networks and devices
Transparency of measurement methods

Background of NetMap
8th Semester project

Mismatch probability modelling


Need for network measurements
Add practical work to very analytical/theoretical project
Simple round trip times (network delay) measured
A lot of media publicity

Background of NetMap
P9 semester project

Further development of concept


Additional measurement
parameters/methods

P10 semester project

Deep analysis of measurement


methods
Lowering data consumption

System Overview
Components in the NetMap system network measuring
process

Devices performing the measurements


Server placement
Connection between device and server

System Overview
Considerations regarding components

Android devices
Server location
The network

Measured Parameters
What we measure

Round Trip Times (RTT)

Response times, Network delay

Throughput

Bandwidth, Connection speed

Handovers

Of connection, Technology, Sub-technology

Connectivity

To the server, Soft performance of network

Measured Parameters
Round Trip Times [ms]

Both TCP and UDP pings


Very small payload

Measured Parameters
Throughput [Mb/s]

TCP
Up-/Download
15 second test

10

Measured Parameters
Handovers [#]

Connection technology and cell tower


Event based

11

Measured Parameters
Connectivity [%]

1 second ping
30 tests pr calculation

12

Measurement Scheduling
Measurement schedule

Simple round robin at fixed interval


Store results on device and transfer when full

Reduced network overhead


Make part of measurement scheduling

13

Measurement Scheduling
Alternative measurement schedule

Goals: reduce resource consumption


Adaptive measurement schedule

Only measure if data is needed in area


Adapt the frequency of measurements accordingly
Other

Measurement collection

Only on WiFi
Only at night
Other

14

Possibilities from Measurements


Parameters can show variations caused by

Location
Environment

15

Possibilities from Measurements


Parameters can show variations caused by

Time of day

16

Possibilities from Measurements


Parameters can show variations caused by

Area type

17

Possibilities from Measurements


Parameters can show variations caused by

Network operator

18

Possibilities from Measurements


Parameters can show variations caused by

Device model

19

Possibilities from Measurements


Parameters can show variations caused by

Movement speed of device

20

End User Involvement


Critical for success

No users => No data

Background
Demographic

Why should a user spend his data on us


What do they have to do
What will they get out of it

Getting all kinds of users to join

Motivation of users

21

Are you ready for some polls?

A. Ja!
B. Yes!
C. Tak!
D. J!
E. Si!
F. WUHUU!
22

SecondsFastest
Participant
Responders
1,19
5F9F59
1,415
5F9F60
1,527
5F9FC2
2,257
5FA08D
2,698
5FA07E
2,82
5FA0AB
3,153
5FA0AC
3,244
5FA093
3,397
5FA0B9
3,665
5FA083

Seconds
3,926
3,953
4,262
4,581
5,247
5,641
5,696
5,788
5,808
5,833

Participant
5FA05C
5FA0A6
5F9F9F
5F9F51
5FA097
5FA0A5
5F9FA8
5FA074
5FA05F
5FA09F
23

Do you own a smartphone with a data connection?

13%

Ye
s

A. Yes
B. No

87%

24

Rate the reasons for why you would be reluctant to


install the NetMap app and perform measurements
starting with the most important

A. I worry about my
privacy
B. I dont want to spend
all my data on this
C. I dont want to spend
my battery power on
this
D. I am not interested in
this area

40%

25%
21%
14%

A.

B.

C.

D.

25

Rate the reasons for how you could be motivated to


install the NetMap app and perform measurements
starting with the most important
A. Possibility to win something
(vouchers, gift cards)
B. Getting free resources in
return (data, minutes)
C. Personal interest in results
(performance of your device)
D. Getting documentation
towards ISP of product
performance
E. Helping make the network a
better place (discovering
coverage and performance
issues)

26%
20%
18%

A.

18%

B.

C.

18%

D.

E.

26

Which of the following possibilities would be the


most useful to you?

A. Being able to
compare device
performance when
buying new devices
B. Being able to
compare ISP network
performance when
choosing subscription
and ISP

55%
45%

A.

B.

27

Final Notes
Many possibilities

In analysis of measurements
With crowd sourcing
Simple network parameters

Featured articles
Conference invite
Interest from industry
Interest from general public

Huge interest

28

Final Notes
Featured articles

29

Final Notes
Crowd sourcing

Attract users with competition


Attention through media

30

Thank you for your attention

31

Ready to go to the beach?


12

A. Yes!
B. No

No

Ye
s!

32

Das könnte Ihnen auch gefallen