28th August 2014. My presentation at SEAA 2014 (http://esd.scienze.univr.it/dsd-seaa-2014) about our survey on the specification of the physical environment of Wireless Sensor Networks (WSNs). Accompanying paper: TO APPEAR Abstract: A wireless Sensor Network (WSN) consists of spatially distributed sensor nodes that cooperate in order to accomplish a specific task. What really sets WSNs apart from all the other kinds of distributed systems is the limited processing capabilities of the nodes, contingent energy restrictions, and their strict dependence to physical phenomena like attenuation, reflection, etc. Under this perspective, the physical environment in which WSN nodes are deployed strongly affects the overall quality of the system. Under this perspective, how WSN engineers currently specify the physical environment and how they would like to do it? This paper presents a survey we run by interviewing WSN engineers with a special focus on their practical needs and activities. By analyzing the collected data, we can conclude that: a) a good number of practitioners describing the physical environment do it by GIS software or informally, b) practitioners not specifying the physical environment do not see a clear return on investment on doing it, c) practitioners rate as (definitely) useful a potential tool for deploying WSN nodes on a virtually specified physical environment.

1. A Survey on
the Specification of the Physical Environment
of Wireless Sensor Networks
Ivano Malavolta
Henry Muccini

2. Roadmap
Background
Contribution
Design of the study
Results
Takeaways
Conclusions

3. Wireless sensor networks (WSNs)
WSNs consist of spatially distributed sensors that cooperate to
accomplish some tasks.
Sensors are:
– small
– battery-powered
– with limited processing power
– with limited memory
They can be easily deployed to monitor different environmental
parameters such as temperature, movement, sound and pollution.

4. WSN applications
Sensors can be distributed on roads, vehicles, hospitals, buildings,
people and enable different applications such as:
• environmental monitoring
• medical services
• battlefield operations
• crisis response
• disaster relief

5. WSN physical environment (1)
What really sets WSNs apart from all the other kinds of distributed
systems is:
• limited processing capabilities of the nodes
• contingent energy restrictions
• strict dependence to physical phenomena
like refraction, reflection, and attenuation…
à The physical environment in which
WSN nodes are deployed strongly
affects the overall quality of the system

6. WSN physical environment (2)
Information from the physical environment, like:
• exact position of the nodes
• information about the surrounding obstacles and their material
– e.g., walls, furniture, windows, or small objects in general
surely helps making an accurate estimate of the physical phenomena
affecting the WSN
Such data could allow a more precise
measurement of the network in
terms of: bit error rate, packets loss,
energy consumption, etc.
à enables the prediction of how the
WSN will globally behave when
nodes are deployed in different ways

7. Examples
VeriSensor [1]
GLONEMO [2]
[3]

8. Roadmap
Background
Contribution
Design of the study
Results
Takeaways
Conclusions

9. Contribution
To investigate on how practitioners specify the physical
environment of a WSN
Survey by interviewing WSN practitioners with a special
focus on their practical needs and activities
• Many practitioners describe the physical environment
via GIS software or informally
• practitioners not specifying the physical environment
do not see a clear return on investment on doing it or
perceive existing algorithms and tools as too complex
• practitioners rate as definitely useful a potential tool
for deploying WSN nodes on a virtual environment
GOAL
HOW
MAIN
FINDINGS

10. Roadmap
Background
Contributions
Design of the study
Results
Takeaways
Conclusions

11. Research objective
Our main research question is
How WSN engineers currently define the physical
environment, and how they would like to do it?
Why they should define it? To better reason on:
• the network topology
• how much power is consumed by the application running on the nodes
with respect to the used batteries or harvested energy sources
• how well an area is covered or tracked by sensors
• …

12. Research sub-questions
Do engineers explicitly specify the physical environment
where the WSN is going to be deployed?
RQ1
If so, how do they accomplish this task (e.g., formally, informally, etc.)?
Do engineers specify the sensor nodes and their exact
position within the physical environment of a WSN?
RQ2
If so, how do they do it (do they consider obstacles, hardware configuration, etc.)?
What are the most relevant features a potential tool for
specifying the physical environment of a WSN shall expose?
RQ3
Need to consider the exact shape of obstacles, or only an approximation?
How would WSN engineers prefer to interact with such a potential tool?

13. Population selection (1)
Participant profile:
Engineer who has been concretely involved in the
development of at least one WSN in the last 10 years
Two sampling methods:
1. Convenience sampling - we directly selected WSN engineers from:
– our personal contacts
– reference websites, newsgroups, and other web resources about WSN
OSs, node vendors, and WSN technologies in general

14. Population selection (2)
2. snowball sampling [4] - we asked selected participants to nominate
additional experts in their network
Resulting population
21 WSN engineers from 18 different organizations in 9 countries
Main affiliation types:
– university
– center of excellence
– company
– research institution
image from: http://www.hsrmethods.org/Glossary/Terms/S/Snowball%20Sampling.aspx

15. Design of the questionnaire*
a.Introduction
b.Personal
information
Yes
Is the WSN
environment
specified?
c. Questions about the
WSN environment
specification
No
c. Questions
about why and
how the WSN
environment is
not specified
Is the WSN
environment
specified
digitally?
Yes
c. Questions about
digital WSN
environment
No
e. Questions about
the potential tool for
WSN environment
d. Questions about
WSN Design
f. Concluding
questions
Yes
Involved in the
WSN design
phase?
No
21
7
close-ended questions
open-ended questions
a) purpose of the study + terminology
b) demographical info of participants
c) how environment is specified
d) focus on nodes and positioning
e) potential tool for WSN environment
f) additional comments + snowballing
A transcript of the questionnare is available here: http://www.di.univaq.it/malavolta/wsn/WSNenv.pdf

16. Roadmap
Background
Contribution
Design of the study
Results
Takeaways
Conclusions

17. Population
21 practitioners:
14 with experience ≥ 5 years
7 with experience < 5 years
1
1
14
2
3
1
1
15
3
1
0
2
4
6
8
10
12
14
16
1000 and above
100-999
50-99
10-49
1-9
Average number of WSN nodes
Number of nodes in the largest WSN project
53%
23%
19%
5%
#projects < 3
3 ≥ #projects ≤ 6
#projects > 6
No info
43%
28%
5%
24%
Equally indoor and outdoor
Mostly indoor
Mostly outdoor
Indoor only

18. WSN environment specification (1)
Encouraging for our study since
we can investigate on both types
of development processes
Major trend in specifying the
environment in a precise way,
rather than relying on draft
specifications.
48%
52%
The WSN environment is
explicitly specified
The WSN environment is
not specified
20%
30%30%
10%
10%
Always by a draft
Mostly by a precise
specification
Equally
Not specififed
Always by a precise
specification

19. WSN environment specification (2)
Clear trend in favor of digital
representation
Most used file formats:
text-based and images
Basically, those results uncover
the great variance about the
software used to represent the
WSN environment
90%
10%
Digital
representation
Paper-based
representation
40%
30%
20%
10%
Maps and GIS
software
Office software
Dedicated
software
Don't know

20. 2D vs 3D
80%
10%
10%
2D
3D
2d and 3D
Due to the complexity of producing
3D models?
Due to the fact that 2D models are
perceived to be sufficient for
representing the environment of a
WSN?
In this case, 2D+3D representation
is the main trend
In their last project
Best options in general?
20%
30%
50%
2D
3D
2d and 3D

21. Obstacles definition
33, 33%
13, 13%
33, 33%
6, 6%
15,
15%
Free space (no obstacles)
Walls, floor, and roof
Walls, floor, roof, windows, and
large-sized objects
Walls, floor, roof, windows,
large and small-sized objects
No choice
Clear winners:
• free-space environment
• only very large obstacles (e.g., walls, roofs, etc.)

22. Hardware and nodes positioning
94%
10%
Definetely useful
Not useful
Indeed, WSN engineers must have
at least some knowledge about the
hardware features of the nodes
used in the WSN.
Examples:
– transmission power of the antenna
– available sensing devices
– batteries voltage
Do analytical models and
simulation tools fit well
with practitioners’ needs?
Usefulness of having a hardware specification
Instrument for evaluating the optimal nodes positioning
84%
0%
16%
By deploying them on
site (real-world testbed)
Analytically
By simulating the
network
Other
“Simulation is performed only if
simple, feasible and meaningful,
otherwise deployment”

23. Why not specifying the WSN environment?
Why not?
54%
46%
No perceived
usefulness
Lack of satisfactory
tools, algorithms or
models
“Because up to now it has been
sufficient just to know the main
features of the environment”
“We mainly worked on networking
protocols, able to adapt to the changes
of the environment”
“Unclear whether the modeling
effort is going to pay off”
How do they proceed to the deployment of the WSN?
37%
27%
27%
9%
Not needed (adaptable
WSN)
Measure the WSN on the
field, after deployment
Preliminary measures of the
area and network simulation
Based on their experience
“It is simpler not to
model the environment
and compensate for
time dynamic failure
with robust algorithms”

24. Potential tool (1)
Proposal: potential tool that allows engineers to virtually deploy a WSN
in the environment.
Such a potential tool could simulate an environment where to virtually
deploy a set of defined sensor nodes into a digital version of its physical
environment.
48%
14%
38%
0%
Definetely useful
Useful
Neutral
Not very useful
Definitely not useful

25. Potential tool (2)
24%
33%
43%
Tool interaction
By importing a file produced by means of an external
tool (for example Autocad)
By directly drawing the environment within the tool
By firstly importing an image file to be used as a guide
to the drawing phase within the tool
When asked about their interest in defining the exact shape of the
obstacles, no clear trend has been identified

26. Potential tool (3)
About the importance of physical effects for the WSN:
weighted sum
Physical effect
-2
-1
0
+1
+2
ws
Attenuation
0
0
0
7
14
7
Reflection
0
0
2
11
8
5.4
Scattering
0
1
6
7
7
4
Diffraction
0
2
6
11
2
2.6
Refraction
0
4
5
9
3
2.2
Polarization
0
4
8
7
2
1.4

27. Roadmap
Background
Contribution
Design of the study
Results
Takeaways
Conclusions

28. Do engineers explicitly specify the physical environment
where the WSN is going to be deployed?
RQ1
Good number of practitioners explicitly define the WSN environment
Almost equal number of practitioners do not
– mainly they do not see a clear ROI
– no satisfactory tool or method
à Researchers should
• provide a more concrete evidence about the advantages of
explicitly representing the WSN environment
• work further on methods, algorithms, and tools
Majority of participants would prefer to
– define the physical environment via mapping or GIS software
– use a combination of text and images
– use a combination of 2D and 3D representations

29. Do engineers specify the sensor nodes and their exact
position within the physical environment of a WSN?
RQ2
WSN practitioners typically:
– consider free-space environment
– consider only very large obstacles (e.g., walls, roofs, etc.)
– rely on physically measured testbeds
à do current simulation and analysis techniques demand too much
effort to WSN practitioners?
“Usually the available simulation tools do not provide a functionality to define
and describe the environment. However, I feel that it is equally important to
describe the environment and its behaviour in addition to the models that
define how the networking part will function. I believe this is due to the
difficulties in defining accurate models for the environment.”

30. What are the most relevant features a potential tool for
specifying the physical environment of a WSN shall expose?
RQ3
WSN practitioners strongly need a tool for:
1. defining the physical environment of a WSN
2. virtually deploying WSN nodes into it
The tool may allow engineers to specify the environment in different ways.
For example, by importing an image that will serve as the basis for a
subsequent drawing phase.
We believe that this option provides a good trade-off in terms of level of
usability and preciseness
Mininal set of physical effects to be considered: attenuation and reflection

31. Roadmap
Background
Contribution
Design of the study
Results
Takeaways
Conclusions

32. Conclusions
“I think that a study on modelling and analysis of the WSN
environment is interesting and can give you some new ideas
because nowadays in most cases a WSN is intended as a set of
hardware nodes, without taking into account the place where the
nodes will be deployed”

33. References
[1] Y. Ben Maissa, F. Kordon, S. Mouline, and Y. Thierry-Mieg, “Modeling and analyzing
wireless sensor networks with verisensor: An integrated workflow,” in Transactions on
Petri Nets and Other Models of Concurrency VIII, ser. Lecture Notes in Computer
Science, M. Koutny, W. Aalst, and A. Yakovlev, Eds. Springer Berlin Heidelberg, 2013, vol.
8100, pp. 24–47. [Online]. Available: http://dx.doi.org/10.1007/978-3-642-40465-8 2
[2] L. Samper, F. Maraninchi, L. Mounier, and L. Mandel, “Glonemo: Global and accurate
formal models for the analysis of ad-hoc sensor networks,” in Proceedings of the First
International Conference on Integrated Internet Ad Hoc and Sensor Networks, ser.
InterSense ’06. New York, NY, USA: ACM, 2006. [Online]. Available: http://doi.acm.org/
10.1145/1142680.1142684
[3] http://www.remcom.com/wireless-insite
[4] B. Kitchenham and S. L. Pfleeger, “Principles of survey research: part 5: populations
and samples,” SIGSOFT Softw. Eng. Notes, vol. 27, pp. 17–20, September 2002.

34. Ivano Malavolta |
Gran Sasso Science Institute
+ 39 380 70 21 600
iivanoo
ivano.malavolta@gssi.infn.it
www.ivanomalavolta.com
Contact