You are on page 1of 5

2014 IEEE 25th International Symposium on Personal, Indoor and Mobile Radio Communications

A Reliable Connectivity Based Node Placement Strategy

in Linear and Hierarchical Wireless Sensor Networks
Salman Ali, Saad Bin Qaisar

Emad A. Felemban

School of Electrical Engineering and Computer Science,

National University of Sciences and Technology, Pakistan
salman.ali, saad.qaisar @ {}

College of Computer and Information System,

Umm Al-Qura University, Mecca, Saudi Arabia
Sensor locations also influence energy usage as it is directly
related to the distance between the transmitter and receiver.
Over-provisioning of resources can be avoided by use of
effectively dimensioning the WSN node resources. Monitoring
of linear and hierarchical infrastructures like oil and gas
pipelines, bridges and tunnels proves much challenging for
being extensively long [2]. Even small interruptions from a
limited portion can disturb measurement accuracy from a
major portion of the network. An intelligent network design
and deployment of WSN nodes can improve performance by
maximizing reliability in terms of link quality and network
connectivity. Network connectivity being essentially linked
with link quality depends upon signal strength between sensor
nodes measured as Signal-to-Noise- Ratio (SNR) or Received
Signal Strength Indicator (RSSI). Network dimensions design
and measurement techniques include path losses, capacity,
link-quality and network coverage for deployment in complex
topologies. This paper aims to provide a relation towards
selection and deployment of sensor node resources to help
designing practical WSNs for field deployment in linear and
hierarchical layout.
Linear and Hierarchical WSN provide a special case of
design topology where the difficulty mainly lies in physical
inspection of sensors as well as long distance multi-hop
communication [2][3][4]. A hierarchical infrastructure network
(Fig. 1) is a constituent of several linear portions arranged in a
tree topology. Independent solutions exist for path loss and
connectivity that mostly rely on MAC protocol for real time
networking however MAC approaches become impractical due
to severe time restrictions. Sensor Network Calculus (SNC)
provides a resource selection mechanism but is limited in terms
of initially fixed resources. Transmit power efficient sensor
placement schemes for linear WSNs focus on minimization of
average energy consumption per node with an objective to
maximize network lifetime [7][9][10]. The limitations to such
an approach lie in defining a proper optimization problem that
is normally not suitable for real time deployments.
A number of sensor node placement approaches also try to
balance between network traffic load and lifetime
maximization [11]. Quite a few methods intentionally focus on
optimal required node density and the spacing in-between to
intelligently use network resources for linear and hierarchical
distributed detection applications addressing scalability and
collaboration but the major concern lies in the complexity of
calculations [12].
Apart from successful event detection in signal processing
domain, path losses in terms of incurred packet errors for low
channel SNR/RSSI severely affect the reliability with which
data can be communicated over the span of the infrastructure
[14]. Accurate channel models can only be determined for
specific layouts of network infrastructure and the terrain type
involved [15][16]. Such issues become significantly important

AbstractWireless Sensor Networks have been used

extensively for environment data collection from dense as well as
simple linear structure deployments. However, the need to collect
data from all sensors in the network imposes constraints on internode distances. Critical issues related to optimal resource
allocation still persist for even simplistic deployments like linear
and hierarchical networks. In this work, we present a method
for node placement in linear wireless sensor network with notion
of provisioning reliable data delivery over extensive monitoring
infrastructures. We provide a selection mechanism for required
number of nodes that can cover a specific area while providing a
margin for channel reliability. The reliability approach has been
analyzed in a pipeline monitoring scenario and subsequently
referred with analytical model to ensure consistency with
minimum resource consumption and best network connectivity in
real world. The proposed node placement strategy is based on a
dynamic programming model for achieving reliability in received
signals in a multi hop and long distance connectivity perspective.
Index Termsdynamic programming, linear network, channel
reliability, wireless sensor network

Advances in embedded design and wireless communication
have led to the development of miniature sized low power
multi functional sensor devices. These devices form the basis
of Wireless Sensor Networks (WSNs) that can monitor diverse
scenarios like battlefield, health and various infrastructures [1].
The sensor device or sensor node is capable of detecting
environment parameters within its sensing range and
transmitting the acquired data to nodes within its
communication range. The sensed data can then be sent in a
multi hop manner to a central sink. Nodes collaborate with
each other to monitor inaccessible areas and to communicate in
tough terrains. WSNs can be categorized in several ways when
considering coverage and connectivity requirements. Coverage
requirements can be the same throughout the monitoring area
mandating uniform node layout while it may be critical in
some other areas with a need for higher surveillance. For
regular infrastructures, a finite set of points can be used for
approximating the coverage requirements with a finite set of
points. Once nodes are deployed, network reliability would
depend upon the inter node connectivity and channel quality
statistics including packet error rates.
For multi-hop communication, it is essential for reliability
that the nodes can communicate with each other at all times.
Network is considered fully connected if every pair of nodes
can communicate with each other either directly or through a
relay node. Connectivity is related to the position of nodes that
is also affected by the channel statistics like SNR (Signal-toNoise Ratio) and RSSI (Received Signal Strength Indicator).
The layout of WSN nodes involves several parameters. Sensor
locations that satisfy the budget requirements and coverage
restrictions to counter sensor failures need to be determined.

978-1-4799-4912-0/14/$31.00 2014 IEEE


for long range WSN monitoring infrastructurres contributing to

an overall less spectral efficient design [16].
Traffic aware relay and slave nodes in tree topology can
also be intelligently placed using methods reelying on mapping
the solution into Euclidian distance space takking monetary cost
as a tuple of coverage parameters [12][13].. This approach is
suitable not only in terms of having low com
mplexity but also in
producing practical and closely optimal results. Hence we use a
to define
similar method based on a low complexity algorithm
node resources for linear and hierarchical struucture monitoring.
In this work, we contribute a node resourrce estimation and
placement method for linear and hierarchiccal WSN topology
relating channel conditions, coverage paraameters and node
resources. In contrast with previous work in
i the domain, we
provide path losses and distance profiling with
w common inter
linked parameters (Section II). With focuss on reliability, a
dynamic programming based [17] node plaacement algorithm
that closely follows channel conditions in term
ms of SNR is used
(Section III).

achieve. Since WSN appliications and monitoring for

infrastructure is mostly ussed in intense terrains and
environment, wireless channell related activities like fading,
shadowing and interference crreate considerable loss in signal
strength. To account this, appropriate
models for WSN
applications have been formulated
individually with
experimentation in different teerrains [3-6]. The basis for all
such models is the inversely prroportional relationship of signal
strength to distance between two sensor nodes with slight
from experimentation.
adjustments in path loss factor predicted
Besides path loss, differennt noise forms experienced in
WSN deployed in industrial environments are also critical.
b stochastic process forms a
Such noise when modeled by
superposition of Additive Whitte Gaussian Noise (AWGN) as a
zero mean Gaussian random distributed
process and impulse
noise in the form of randomly distributed
variable [14][16]. We
jointly define such noise forms as:
1,2, , T
x t k t t
where (t) and k(t) are zero mean
Gaussian random variables
and (t)
specifically denotees AWGN, while x(t) being a
binary variable can take on valuues [0,1]. The WSN channel can
be modeled to move between good
and bad states according to
a two-state Markov process (Fig. 3), to describe bursty nature
of impulse noise.
t probability of moving from
If we represent Pr_GB as the
good to bad state then Pr_BG would
be the probability to move
bad to good state. The two staates of the WSN channel can be
represented as [s(t)=G x(t)=
=0] and [s(t)=B x(t)=1]. The
pdf of the stochastic noise in thhe good and bad states can then
be defined through Gaussian vaariable definition as:
|n t |
Pr t |s t
|n t |
Pr t |s t

Average noisee power in bad state
Average noise power in good state
The parameter denotes the standard deviation of noise. For
accurate detection of bad statee, R should have a value greater
than 1, i.e. noise power measuured in bad state must be greater
than any noise power experiennced in the good state. From the
Markov channel state modell, the probability of having a
particular state at any time instaant (t) can be written as:

Fig 1. An example of hierarchical linear infrastructure monitoring with WSN


A major concern in deployment of WSN
Ns for large scale
infrastructure monitoring is the dimensioninng of sensor node
resources in order to ensure reliable annd cost effective
operation. Provision for reliable monitorring is done by
maintaining a tradeoff between link SNR and the network
coverage in terms of transmission distancee corresponding to
RSSI values where data or message can be decoded correctly
with minimum errors. The system model contributes error
probabilities, path loss profiling, network connectivity and
inter-node distances. The symbols used in the paper are
summarized in Table 1.
Aggregation rate






Nominal energy


AWGN variable

Number of aggreegators

Channel bandwidth

Number of hops

Channel capacity

Number of sensoors

Channel constant

Number of transsmissions


Channel loss

Reception Power

Channel state

Single sided noisse

Data bits

Signal Strength

Data rate

Standard deviation

Inter-node distance

Transmission disstance

Link reliability


Time instant
Time period
Transmission poower

Coverage length
Noise at instant i

Pr S 1 T Pr s t 1 |s t
Ps t 1
i|s t
The node separation disttance and path loss derive the
transmit power required to maintain a quality link in
connection with the sensitivityy of used antenna. Free space
model must be adjusted with specifics of path loss exponent
environment. A log-normal
and channel conditions to fit WSN
path loss alteration in the basicc free space path loss model [18]
needs to be integrated in orderr to provide for the accuracy in
loss measures for WSN in near ground outdoor environment.
The path loss as a log-normal equation can be written as:
Pr S t




where is log normal path looss, is path loss at a reference

distance, u is path loss factor and X is a log normal variable
with standard deviation of in dB [3][5][6]. In a normal setting,
may be taken as 36dB, u equual to 4 and X has a variation of
4.70. To compare theorettical path loss formulations,

A. Path Loss Model

For WSN, the transmission rate andd antenna power
critically effect the distance a sensor nodee transmission can







Power: -2dBm
Power: 2dBm









Power: -8dBm
Power: -2dBm
Power: 2dBm




Distance (meters)


RSSI (dBm)

RSSI (dBm)


Distance (meters)



RSSI (dBm)

Distance (meters)










Power: -8dBm
Power: -2dBm
Power: 2dBm






Fig. 2. (a) RSSI of XBee (S2) with Libelium Wasmpotes in Indoor Environment (b) in Outdoor Environment (c) placed over Pipeline

utilized in transmitting and receiving packet from a branch

node in the tree structure of connected nodes n
1 in time T [5][14][15]. This derives the network
lifetime and signal strength as:

experiments were performed using Libelium Waspmotes

equipped with Xbee, Zigbee protocol enabled transceivers
equipped with 2dBi omni-directional antennas. Tests were
conducted for indoor, outdoor (freespace) and linear pipeline
infrastructure of 8inches diameter. The RSSI variations against
distance are provided in Fig. 2. (a), (b) and (c). Interestingly,
the pipeline infrastructure presents similar or improved RSSI
for linear applications since a variation of 2dBm is observed
when compared with normal freespace deployment. The reason
for this phenomenon can be contributed to the superposition of
signals at certain points reflected from the linear pipeline
structure when the nodes are placed above the metal structure.
This however, would be quite different as compared to the
situation where the metal pipeline structure is in the middle of
two nodes causing absorption or blocking of signals.

Power Regeneration Rate >

In mathematical form, we can write (eqn. 10) as

ER . T

conversion to an equation form, we get


. By

Considering the basic relation between transmitted power and

, we can write the fundamental
received power P
relationship between capacity, bandwidth and path loss as:
log 1
By rearranging equation (14), we get the distance at which the
signal can be received effectively by nodes (eqn. 9) as:






B 2B


A . T. E . u. n
. T. E . b A
. b. D . n

is the number of sensors deployed. For a typical
WSN, optimal distance placement achieves a reliable link
under the constraint of maximum lifetime as a function of
average and initial energy. But the nodes are placed at the
minimum tolerable SNR region boundary where any slight
displacement will lead to disconnectivity which we cater in this
work using a dynamic programming based node placement
algorithm. The optimal distance placement is done by
maximization of lifetime as a function of average and initial
energy as [7]:

aD R b R


. T. E

ER , E
and E
are the energy regeneration rate, signal
transmission and amplification energy respectively while
is the number of sensors connected to the aggregator in
is the aggregation rate and b is the number
a tree branch. A
of data bits transmitted. Aggregation rate refers to the data rate
that can be received from several branch nodes over a time
period T. Alternatively; it can be represented as a percentage
ratio in terms of maximum data rate (250kbs) that can be
received from a single node in one unit time. The maximum
tolerable SNR distance depends upon the discrete transmission
capability of the node; hence sensor i would select a discrete
value P where j, in the case of our experimental setup with
Libelium Waspmotes, increases in six steps to a maximum of
1mW. In the most simplistic linear case for equal distance
placement, the distance between adjacent nodes will be
where L is the network length
adjusted as D

B. Transmission Distance Calculation

Path loss and channel characteristics determine the
transmission distance at which sensor nodes should be placed
apart for maximum throughput. It should be noted that the
transmission range has variations for an omni-directional
antenna. Considering this, there is always an SNR gap for a
shift from a good reliable connection to a bad connection
where the packet reception may suffer losses. Hence we derive
several measures of inter-node distance placement.
Considering power received to be proportional to the ratios of
distances where the receiver is present and some relative
distance at which loss is measured, we have P

Subject to,
By using Lagrangian multiplier method,
, 1 i n 12

Here, u is the path loss component that intrinsically relates to
reliability in terms of SNR. A heuristic based approach with
notion of reliability can also be used instead of the optimal
placement since nodes can undergo disconnection for being

For best accuracy in measurement, the path loss exponent u can

be estimated directly from the log-normal utility as :

log D
Maximum distance where SNR is minimum but signal can
still be decoded presents the transmission distance after which
the signal will drastically get altered by interference. This
maximum tolerable SNR region can be derived by setting the
energy regeneration rate greater than or equal to the energy


placed on the boundary of transmission region. The heuristic

method scales the distance as a function of the SNR reliability
achieved in reducing the distance between nodes and the
number of budget nodes that can be accommodated. The node
placement distance is:
D D _
is the path loss catered effective distance and D is
D _
a scaling factor for coverage determined by dynamic
programming discussed in Section III.
Fig 3 (a)

C. Linear and Hierarchical WSN Resource Requirement

Number of sensor nodes deployed for infrastructure
monitoring constitutes the main resource and cost of WSN.
Hence, a critical and resourceful measure is required for
practical deployment of nodes. From the distance calculations
(eqn. 9), it follows that the number of optimal nodes required
would be given as in [7]:


Coverage Algorithm


where ,



If Total Covered distance <

8. Check constraints

is maximum sensing range taken equal to the transmission
range here. It follows that n node
the number of nodes should not exceed the node budget.

9. If no constraint in step 8. met,

Repeat step 4 - 8
10. Else Exit
11. Report current SNR/Spectral Efficiency (db) gain


Dynamic programming [27] is to provide tradeoff between
coverage and node resources utilized against the SNR and
corresponding reliability gain. The problem here is to find the
portion of coverage in transmission range that the node can be
placed inside while meeting the budget nodes, i.e. maximum
number of nodes that can be deployed. With representation of
channel reliability in terms of SNR as S, dynamic algorithm
based reliability oriented node placement and coverage
approach is explained (Fig. 3. (a)).

The population size of dynamic algorithm is also defined

which determines the number of calculations to make at each
step. The starting reliability S is thus set as the minimum
achievable SNR. A small decrease in distance is calculated and
the corresponding SNR gain is calculated. For each change in
SNR and distance, the minimum of their ratios is taken in a
population. The algorithm continues until a constraint in terms
of maximum nodes that can be deployed, maximum SNR or
minimum node separation is met. During the algorithm sorting,
whenever the infrastructure coverage becomes short, a node is
deployed to suffice. At the end of the algorithm, the spectral
efficiency is reported which depicts a sufficient reliability gap.

Fig. 3. (a) Dynamic programming for reliable connectivity Fig 3 (b) Optimal,
dynamic algorithm based and geometric placement of sensor nodes with
reference to transmission distance (r)




Spectral Efficiency


(Dynamic) Avg SNR=10dB

(Dynamic) Avg SNR=5dB
(Dynamic) Avg SNR=0dB
(Dynamic) Avg SNR=-5dB
(Optimal) Avg SNR=-10dB








10 12 14 16 18 20

Number of Branch Node



Number of Nodes

Subject to,

Distance (meter)

Set Coverage Length

2. Define dynamic programming Population Size Pop

3. Initialize starting reliability
(minimum achievable SNR)
corresponding to maximum transmission distance
4. Evaluate a population with decrease (D) (meters) in distance
and corresponding increase in (S) (dB)
5. Set
S (Relative change in SNR) for all deployed nodes
6. For each (S, D) pair from population, evaluate

arg max T

arg max

Fig 3 (b)


Optimal Palcement
Dynamic Placement
(2/3)r Placement
(1/2)r Placement


8 10 12 14 16 18 20 22

Number of Nodes







Fig. 4. (a) Node distance variations with channel index u (b) Spectral efficiency gain for number of branch nodes and average SNR (c) Node Resource
Comparison for dynamic algorithm approach against optimal and geometric placement




This work is funded by grant number NSTIP-10-ELE1238-10

from the Long-Term National Plan for Science, Technology
and Innovation (LT-NPSTI), the King Abdul-Aziz City for
Science and Technology (KACST), Kingdom of Saudi Arabia.
We thank the Science and Technology Unit at Umm A-Qura
University for their continued logistics support.

We compare our dynamic algorithm strategy against

optimal placement at the transmission range (r), and
geometrical placements [20] at (2/3)r and (1/2)r inside the
signal footprint of an omni-directional Zigbee antenna and a
test bed consisting of Libelium Waspmotes (Fig. 3 (b)) [19].
For linear and hierarchical networks with difficult and lengthy
terrains, determining the precise channel index to use presents
precision difficulty. Though much distance change in terms of
transmission coverage for sensing nodes does not provide
considerable gains and as the node index increases the node
distance for the same infrastructure length decreases and the
role of the channel index does not provide considerable
difference in channel quality since overall the average SNR
and spectral density increases (Fig. 4 (a)).
Major gain achieved from the use of dynamic programming
is the spectral efficiency increase for the number of nodes
deployed to cover the same infrastructure length but with
varying distance against average SNR (Fig. 4 (b)). The optimal
placement is done for a minimum SNR of -10dB and the gain
against the placement with reduced distance and improvement
in SNR is plotted. Considerable spectral efficiency gains in the
range [0.1-2.5] and more can be achieved depending upon the
SNR gap for channel conditions and node placement. It is more
profound for dynamic algorithm at earlier stages when linearly
connected nodes or the number of hops is limited in the range
{1-15} on average. This is because as the same packet travels
over multiple hops each having a slight probability of error; the
overall spectral efficiency will reduce.
The node resource consumption for dynamic algorithm is
slightly more than the optimal approach since a compromise is
made the increased packet reception rate and improved channel
quality measured with SNR and spectral efficiency (Fig. 4 (c)).
The resource consumption is however less in the order and
for geometrical placements at (2/3)r and (1/2)r. To summarize,
the dynamic algorithm proposed can be used as a tool to
alternatively deploy nodes in a more reliable and connected
manner within resource constraints while benefiting from
channel efficiency gains.

[1] Franck L. Lewis, "Wireless sensor networks" Smart Environments:
Technologies, Protocols, and Applications, pp. 11-46, 2005.
[2] Z. Wang, X. Zhao, X. Qian, "The application and issues of linear
wireless sensor networks," International Conference on,
Engineering Design and Manufacturing Information, pp. 9-12, 2011.
[3] C. W. Chen, Y. Wang, Chain-Type Wireless Sensor Network for
Monitoring Long Range Infrastructures: Architecture and
Protocols, Hindawi IJDSN, vol. 4, issue 4, pp. 287-314 4, 2008.
[4] I. Jawhar., N. Mohamed, A hierarchical and topological
classification of linear sensor networks, Proceedings of the
Wireless Telecommunications Symposium, pp. 72-79, 2009.
[5] I. Jawhar, N. Mohamed, K. Shuaib, N. Kesserwan, An Efficient
Framework and Networking Protocol for Linear Wireless Sensor
Networks, Ad Hoc & Sensor Wireless Networks, pp. 119, 2009.
[6] J. B. Schmitt, U. Roedig, Worst case dimensioning of wireless
sensor networks under uncertain topologies, IEEE Proceedings of
Resource Allocation in Wireless Networks, 2005.
[7] M. Cao, L. T. Yang, X. Chen, N. Xiong, Node Placement of Linear
Wireless Multimedia Sensor Networks for Maximum Network
Lifetime, Advances in Grid and Pervasive Computing, vol. 5036,
pp. 373-383, 2008.
[8] M. Cao, T. Laurence, Yang; X. Chen, N. Xiong, Node Placement
of Linear Wireless Multimedia Sensor Networks for Maximum
Network Lifetime, Proceedings of the 3rd international conference
on Advances in grid and pervasive computing, pp. 373-383, 2008.
[9] Cheng. P.; Chuah, C. Nee, X. Liu, "Energy-aware node placement
in wireless sensor networks," IEEE Global Telecommunications
Conference, vol. 5, pp.3210-3214, 2004.
[10] M. Esseghir, G. Pujolle, Wireless Sensor Nodes Dimensioning
under Network Lifetime Constraint, IEEE 66th Vehicular
Technology Conference, pp. 76-80, 2007.
[11] H. Li, X. Shunjie, "Energy-Efficient Node Placement in Linear
Wireless Sensor Networks," International Conference on Measuring
Technology and Mechatronics, vol. 2, pp.104-107, 2010.
[12] X. Li, D. K. Hunter, S. Zuyev, Coverage Properties of the Target
Area in Wireless Sensor Networks, IEEE Transactions on
Information Theory, vol. 58, no.1, 2012.
[13] S. Bandyopadhyay, E. Coyle, An energy-efficient hierarchical
clustering algorithm for wireless sensor networks, IEEE ICC, vol.
3, pp. 1713-1723, 2003.
[14] G. Zhou, T. He, S. Krishnamurthy, J. A. Stankovic, Models and
solutions for radio irregularity in wireless sensor networks, ACM
Transactions on Sensor Networks, vol. 2, pp. 221-262, 2006.
[15] A. Martinez-Sala, J. M. Molina-Garcia-Pardo, E. Egea-Ldpez, J.
Vales-Alonso, L. Juan-Llacer, J. Garcia-Haro, "An accurate radio
channel model for wireless sensor networks simulation," Journal of
Communications and Networks, vol. 7, no.4, pp.401,407, 2005.
[16] M. Cheffena, Industrial wireless sensor networks: channel
modeling and performance evaluation, EURASIP Journal on
Wireless Communications and Networking, 2012.
[17] J. L. Williams, J.L, J. W. Fisher, A. S. Willsky, "Approximate
Dynamic Programming for Communication-Constrained Sensor
Network Management," IEEE Transactions on Signal Processing,
vol.55, no.8, pp.4300-4311, 2007.
[18] S. Kumar, D. K. Lobiyal, "Linear order sensor network deployment
for coverage analysis based on TBM theory," IEEE 3rd
International Advance Computing Conference, pp.404-407, 2013.
[19] M. Younis.; K. Akkaya, Strategies and techniques for node
placement in wireless sensor networks: A Survey, Elsevier Ad Hoc
Networks, vol. 6, issue. 4, pp. 621-655, 2008.

The paper presents a dynamic algorithm based framework
for sensor node resource estimation and placement in WSNs
for reliably monitoring in linear and hierarchical cases. An
analytical foundation based on path loss and distance profile is
used to calculate the required resources for full coverage. Test
bed deployment consisting of ZigBee based modules and
comparison for transmission range based geometrical
placement of nodes against proposed dynamic algorithm
assisted node placement strategy has been used to ensure
minimum resource consumption and maximum network
connectivity. Results illustrate that the algorithm prediction for
connected network setup are reasonably close to the actually
observed values in the simulations and testbed for a linear
topology in terms of path losses, signal strength and distance
between nodes. Most importantly for critical reliable
applications, the connectivity challenge is met for major
deployed topology cases. The real time test bed deployment
results together with theoretical networking foundations can be
used as a promising candidate for consistent deployment of
sensor nodes in random and massive quantity for applications
like structural health monitoring with multi hop and long
distance connectivity.