POWER: Planning and Deployment Platform for Wireless Sensor

POWER: Planning and Deployment Platform for Wireless Sensor Networks *
Jinghao Li1, Yuebin Bai1, Haixing Ji1, Jihong Ma2, Yong Tian1 and Depei Qian1
1
School of Computer, Beihang University, Beijing, 100083, China
2
Handan Polytechnic College, Handan 056000, China
[email protected]
Abstract
Impeded by the bottleneck of the information
collection in the information chain, researchers are
more and more interested in the development of
wireless sensor networks; the applications of wireless
sensor networks are growing too. But less attention are
paid to wireless sensor networks planning. To resolve
deployment challenges, reduce the deployment risk, we
propose to build the planning and deployment platform
for wireless sensor networks. POWER is a software
environment for planning and deploying wireless
sensor networks applications into actual environment.
POWER has three main parts, network deployment,
simulation, and performance evaluation and
optimization. Simulation is the foundation of the
POWER. Lifetime is the most important performance
evaluation metrics. The target of the POWER is to
supply an integrated and optimal deployment solution
for an actual application.
1. Introduction
With the requirement of the wireless sensor
networks (WSNs) growing, the applications of the
WSNs get more and more. Currently, WSNs have been
widely used in the fields like habitat monitoring,
health-care, smart home, industries, and military [1, 2].
When users deploy the above mentioned WSN
applications into an actual environment, we found that
there are various challenges and problems; thereby we
propose the planning and deployment platform for
WSNs, namely POWER. And, the goal of the POWER
is to:
• support WSNs deployment solution
• reduce the deployment cost and improve the
efficiency of deployment
• afford the whole network performance
evaluation
*
• test new routing protocols and MAC protocols
• accelerate the WSNs practicality
There are limited literature on planning and
deployment for WSNs, but a lot on the WSN
simulator, which is the base of POWER. J-Sim [3], Ns2 [4] and OPNET [5] are the popular widely adopted
network simulators. SENS [6] and EmStar [7] are
application-oriented simulators. TOSSIM [8] is a
discrete-event simulator for TinyOS.
In this paper we propose the problems of planning
actual applications into environment, and give our
solution to solve these problems, which is to build the
POWER. Moreover, we also introduce the
implementation framework of POWER prototype,
which consists of three main parts, network
deployment, simulation, and performance evaluation
and optimization. The framework is based on the opensource J-Sim simulator.
This paper is organized as the following. Section 2
describes the design considerations. Section 3 presents
the POWER framework. Finally, Section 4 concludes
this paper and presents the future work.
2. Design Considerations
2.1. The Problems
To understand the requirements when users deploy
sensors into actual environment, these key problems
can be divided into following five parts.
A. How to place sensor nodes
The placement is divided into two kinds, predetermined or stochastically. In a pre-determined
network, sensors can be placed according to the user’s
request (i.e. manually). In a more realistic WSN
environment, sensors are placed in a stochastically
Project 90612004 and project 90412011 supported by National Natural Science Foundation of China.
Proceedings of the Fifth International Conference on
Grid and Cooperative Computing Workshops (GCCW'06)
0-7695-2695-0/06 $20.00 © 2006
fashion. For example, a common method to distribute
sensor nodes is to scatter them from the plane.
B. The connectivity of WSN
Suppose that every node has a fixed radio range. The
network topology is seen as a mathematic graph, the
target is to studying the graph’s connectivity, i.e. to
determine whether all the sensor nodes can deliver
packets to SINK nodes. In addition, the relationship
between the number of the sensor nodes and the
connectivity; the radio range of the sensor node and the
connectivity should also be studied. In this way, this
can help users to choose optimal sensor nodes and save
the cost.
From above motioned, we found
should contain relative functions to
problems. But these problems are not
they are often interrelated. We have
implement these operations. Figure 1
circulatory process.
that POWER
resolve these
independence,
to circulatory
describes this
C. The coverage of WSN
Coverage which by definition is how well a sensing
field is monitored or tracked by sensors [9] is another
key point. Generally, the coverage validation can
divide into two kinds. One is to confirm how well
some key field is monitored. Another one is in some
application specific scenarios, WSN require a predetermined sensor nodes density in whole areas. All
the two kinds would require extra sensors to be placed,
and also need to modify MAC or routing protocol to
deal with the heavy traffic.
D. How to collect the data
Until now, several kinds of routing protocols and
MAC protocols have been studied. They all base on
different applications background. So, in order to
maximize performance in application specific WSN,
one must not only take into account the routing
protocol but also the MAC protocol. The choice of
MAC protocol or routing protocol can help to
determine the node and the application specific WSN
parameter. Moreover, whether the WSN needs the data
aggregation, including average, minimum, or
maximum data, should also be taken into account.
E. How to evaluate the WSN
Having discussed possible questions and choices,
one must also determine the appropriate performance
metrics. The key metrics for evaluating sensor network
protocols as a whole are lifetime, coverage, latency,
fault-tolerance, scalability, and security [2, 10]. Due to
the character of the WSN, these metrics are often
interrelated.
2.2. Our Solution
Proceedings of the Fifth International Conference on
Grid and Cooperative Computing Workshops (GCCW'06)
0-7695-2695-0/06 $20.00 © 2006
Fig. 1. The implement process of POWER
In figure 1, the first step of POWER is pre-placement.
it manually or stochastically places some sensor nodes
into the virtual environment, and become an integrated
network topology graph. Of course, it should initialize
some WSN parameter, including the nodes parameter,
MAC protocol, and the routing protocol etc. Except
these, it also accomplishes coverage validation and
connectivity validation. The coverage validation is to
determine whether the WSN satisfies the density
requirement and whether the number of sensor nodes is
enough. The connectivity validation is to determine
whether the WSN is connective. Simulation and
Evaluation accomplishes the network simulation and
performance evaluation. When the evaluation is done,
where should be modify to get optimal performance is
the purpose of the optimization. Maybe the
performance bottleneck is the unfitted protocol,
unfitted placement, even the unfitted nodes. When
modified scheme is determined, the process goes back
to the pre-placement to modify placement or to set
models over again. The whole process is a circulation
and the target is to get optimal deployment solution.
The major aim of the POWER is to rebuild the
actual WSN application in the virtual environment, and
supplies a deployment solution. Thus simulation is the
foundation of POWER. We should build several kinds
of models to ensure the accuracy of the simulation. Just
like in the traditional network architecture, the bottom
three layers of the sensor network architecture are
mainly concerned by POWER, which are shown in the
Figure 2. By combining these models, POWER can set
up a virtual network according to the actual network.
course, it should have a drag-and-draw graphics user
interface (GUI) to help user to operate the network
topology.
Moreover, the network topology structure should
also guarantee the requirement of the coverage and
connectivity. Coverage validation helps users to
determine whether the number of the nodes is enough,
whether the nodes density and coverage are satisfied,
where should be added nodes, and where can be
reduced nodes. Connectivity validation is to determine
the network topology graph connectivity.
3.2. Simulation
Fig. 2. The main type of models
3. Overview of the Proposed Framework
This section describes the conceptual framework
design of POWER prototype. According to the above
mentioned
deployment
problems
and
the
implementation process, we found POWER should
contain three main parts. And the key part of POWER
is the simulation. The framework of POWER is shown
in Figure 3. In the following sections, we will describe
these parts in details.
To support WSN deployment solution for a real
application, it is needed to simulate the WSN and carry
out the quantitative analysis of the WSN. So
simulation is the fundamental component of the
architecture.
J-Sim is chosen as the simulator for POWER. J-Sim
is an open-source, component-based compositional
network simulation environment that is developed
entirely in Java. J-Sim was chosen due to its looselycoupled, component-based programming model, as
well as its completed Sensor Network packets [12, 13].
J-Sim includes several kinds of WSN models, such
as radio models, energy models, MAC protocol
models, and routing protocol models. But not all
models are suited for special WSN simulation. To
improve the accuracy of simulation, POWER should
add some new special WSN models, which is the most
important part of this platform. The extensions are
shown as follows.
A.
Fig. 3. The framework of POWER
Protocol models
Just as figure 3 shows, POWER adds new MAC
protocols; they are IEEE 802.15.4, TDMA, S-MAC,
etc. POWER also adds some new special WSN routing
protocols according to different network characteristics
[10]. They are One-Hop, Multi-Hop, and Hierarchical
(LEACH) [14]. Other well known routing protocols,
for example SPEED [15], may be added later.
3.1. Network Deployment
B.
The network deployment aims to find an optimal
placement solution. It contains three main parts, preplacement, coverage validation and connectivity
validation. The pre-placement finishes the placement
of the sensor nodes. Users can manually or
stochastically place each node. Now, quite a lot
literature has studies it; in [11], the author proposes
three typical types of stochastic sensor placement. Of
All the previous preparation work does not help
POWER support environment factor (i.e. assume that
there are no obstructions in the environment), so the
radio propagation model is needed to ensure the
accuracy of simulations. Radio propagation models
attempt to predict the received signal strength at a
given distance from the transmitter. If the strength is
more than a threshold, the sensor begins to receive the
packet. In addition, there are three main phenomena
Proceedings of the Fifth International Conference on
Grid and Cooperative Computing Workshops (GCCW'06)
0-7695-2695-0/06 $20.00 © 2006
Radio models
that affect wireless communication which should be
taken into account. A standard model used to simulate
a clear, unobstructed line-to-sight path between two
nodes is the Friss free-space equation [10]. More
accurate radio model also will be added later.
C.
Energy models
The most important factor in a WSN application is
the energy, as well as the energy model in the
architecture. J-Sim energy model is too simple, which
should be extended. The energy consumption of a
sensor node is divided into three parts: CPU energy
consumption, sense energy consumption, and radio
energy consumption. Thanks to the development of the
micro-electro-mechanical systems, the sense energy
consumption and the CPU energy consumption is very
little. So in the framework, the CPU energy
consumption and the radio energy consumption are
mainly considered. A sensor’s CPU can be in one of
the following four states: active, idle, sleep and off.
Similarly, a sensor’s radio also has four states:
transmit, receive, idle, and sleep. There are two
methods to calculate the energy consumption. The first
method is to assign every state a constant value. When
there is transferring into a different state, relevant value
will be subtracted from the sensor remnant energy with
relevant value. The second method is to calculate
energy according to some equations. For example in
the radio transmit state, energy consumption is in
association with the distant between the two nodes.
The longer the distant, the more the energy
consumption is. This is much flexible than the first
one.
3.3. Performance Evaluation and Optimization
The evaluation part focuses on the evaluations of the
virtual WSN. The appropriate performance metrics
should be determined initially. POWER evaluates the
WSN by following performance metrics: lifetime,
latency, scalability, and reliability. They are used to
determine which protocols as a whole provides the
most efficient paradigm.
Lifetime is the most crucial indicator of the
usefulness of a WSN. Supposing that lifetime is the
time between sensor nodes begins to sense and the
time SINK nodes can get packet from any sensor
nodes.
Latency was the travel time between the packet
creation at a sensor node and the received time at the
SINK node. In some disaster forecast applications, the
WSN must update the critical information in time.
Hence keeping the delay time as short as possible is a
must.
Proceedings of the Fifth International Conference on
Grid and Cooperative Computing Workshops (GCCW'06)
0-7695-2695-0/06 $20.00 © 2006
Reliability means, any protocol that is chosen must
be robust. The dropped packets and error packets
should be as few as possible. The dropped package can
be calculated by comparing the number of packets that
were sent by sensor nodes and the number of packets
that were received by SINK nodes.
Scalability for WSN means the performance is
independent of the number of sensor nodes. It can be
calculated by comparing the performance with
different amount of sensor nodes.
Other performance metrics, including security and
fault-tolerance, are not easy to scale. They will not be
taken into account for now.
Optimization works on the optimization of the
virtual network. According to the evaluation result, we
can find the bottleneck of the performance. Therefore,
we can correct the problem to eliminate or avoid the
bottleneck. Possible solutions include the changing of
the types of sensor nodes, protocols, or some nodes
position, etc. Of course, we can validate the
modification by re-simulate the new WSN according to
the implementation process of POWER.
4. Conclusions
POWER is a software environment to plan the
deployment of the WSN applications. Its target is to
resolve the deployment problems, reduce the
deployment risk, and get an optimal deployment
solution.
In this paper, we analyze the deployment challenges,
and then propose the planning and deployment
platform for WSN, namely POWER, to resolve these
problems. Moreover, we introduce the implement
framework of the POWER prototype. It offers an
integrated process for planning an actual deployment.
Recently, according to the framework, we have built a
prototype of POWER. It only has two types of routing
protocol, two types of MAC protocol, an energy model
and a simple radio model. However it proves that
POWER is credible and useful.
The more accurate models the proposed models are;
the more accuracy the POWER result would be.
Therefore, in future work, new models that POWER
needs would be investigated, such as environment
models, obstacle models, new special protocols, and
new radio models etc. Except these, we also should
take into account about other deployment problems to
improve the reliable of POWER.
5. References
[1] Deborah Estrin, Ramesh Govindan, John Heidemann,
and Satish Kumar.: Next century challenges: scalable
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
coordination in sensor networks. ACM/IEEE
International Conference on Mobile Computing and
Networking archive, ACM Press, Seattle, Washington,
United States, (1999), pp. 263-270.
Sameer Tilak, Nael Abu-Ghazaleh, and Wendi
heizelman.: A taxonomy of wireless micro-sensor
network
models.
Mobile
Computing
and
Communications Review, (2002), pp. 28-36,.
J-Sim Homepage. http://www.j-sim.org. (2005).
Ns-2 Homepage. http://www.isi.edu/nsnam/ns/.
OPNET Homepage. http://www.opnet.com.
Sameer Sundresh, Wooyoung kim, and Gul Agha.:
SENS: a sensor, environment and network simulator.
Simulation Symposium. (2004), 221-228.
J. Elson, S. Bien, N. Busek, V.Bychkovskiy, A. Cerpa,
D. Ganesan, L. Girod, B.Greenstein, T. Schoellhammer,
T.Stathopoulos, and D. Estrin.: Emstar: An
Environment for Developing Wireless Embedded
Systems Software. Technical report, Center for
Embeded Networked Sensoring, University of
California, Los Angeless, (2003).
P. Levis, N. Lee, M. Welsh, and D.Culler.: TOSSIM:
Accurate and Scalable Simulation of Entire TinyOS
Application. In Proceedings of the First ACM
Conference on Embedded Networked Sensor System
(SenSys), (2003).
Chi-Fu huang and Yu-Chee Tseng.: The coverage
problem in a wirless sensor network. Proceedings of the
Workshop on Wireless Sensor Network and
Proceedings of the Fifth International Conference on
Grid and Cooperative Computing Workshops (GCCW'06)
0-7695-2695-0/06 $20.00 © 2006
[10]
[11]
[12]
[13]
[14]
Application, ACM Press, San Diego, CA, USA, (2003),
pp. 115-121.
Nicholas Merizzi.: Sensor Network Deployment in the
McMaster Nuclear Reactor, Master thesis, McMaster
University, (2005).
Ishizuka. M, Aida. M.: Performance study of node
placement in sensor networks. Proc. IEEE ICDCSW’04,
(2004), pp. 598-603.
Ahmed Sobeih, Wei-Peng Chen, Jennifer C.Hou, LuChuan Kung, Ning Li, Hyuk Lim, Hung-Ying Yyan,
and honghai Zhang.: J-Sim: A simulation and emulation
environment for wireless sensor networks. http://www.jsim.org/v1.3/sensor/JSim.pdf, (2005).
Sung Park, Andreas Savvides, and Mani B. Srivastava.:
SensorSim: A Simulation Framework for Sensor
Networks. Proceeding of the 3rd ACM international
workshop on Modeling, analysis and simulation of
wireless and mobile systems, Boston, MA, (2000).
W. B. Heinzelman, A. P. Chandrakasan, and H.
Balakrishnan.: An Application-Specific Protocol
Architecture for Wireless Microsensor Networks. IEEE
Trans. Wireless Communications, Oct. (2002), pp 660670.
[15] He T, Stankovic J A, Lu C, Abdelzaher T F.:
SPEED: A stateless protocol for real-time
communication in sensor networks. In Proc 23rd
Int’l Conf on Distributed Computing Systems,
Providence, Rhode Island. (2003).