Abstract— the routing in wireless sensor networks (WSNs) is
highly critical that increases the latency and congestion. As a
result, throughput performance of the network reduces. In this
paper, scalable mobility-aware pheromone termite (PT) analytical
model is proposed to provide robust and faster routing for
improved throughput and minimum latency. PT also provides the
support for network scalability and mobility of the nodes. The
monitoring process of PT analytical model is based on two
different parameters: packet generation rate and pheromone
sensitivity for single and multiple links.
PT routing model is integrated with Boarder node medium
access control (BN-MAC) protocol. We here also deploy two other
known routing protocols with BN-MAC: sensor protocols for
information via negotiation (SPIN) and Energy aware routing
protocol (EAP). To demonstrate the strength of the PT model, we
have used ns-2.35-RC7 and compared its Quality of Service (QoS)
features with competing routing protocols. The simulation results
demonstrate that the PT is scalable and mobility-aware protocol
that saves energy resources and achieves high throughput by
reducing number of control packets using the BN-MAC as
compared with other routing protocols.
Index Terms— Pheromone Termite (PT) Model, Boarder Node
Medium Access Control (MAC) Protocol, Mobility, Routing,
packet generation rate and pheromone sensitivity.
Wireless Sensor Networks consist of the small size of
sensor nodes. Each sensor node works as a unit with sensing
capabilities to collect and process the data for achieving the
combined goal. The sensors are deployed to observe the
activities of events in the intended areas of interest .
Therefore, it is important to be introduced the system that
should be an energy efficient at all the levels of protocol stack.
The efficient MAC protocol substantially improves the WSN
performance because sensor node consumes enough energy for
accessing the channel. The channel access process is performed
by MAC protocol . The MAC protocol inherits its features
from two major mechanisms: contention and scheduled based
. Contention based MAC protocols are simple and easier to
use without synchronization. However, each sensor in
contention MAC protocol keeps its radio on for a longer period
Manuscript received February 12, 2014 for ASEE Conference.
Abdul Razaque is with the Computer Science & Engineering Department,
University of Bridgeport, CT-06604, USA; phone: 917-889-5975; fax: 917-889-5975;
2 Khaled Elleithy is with the Computer Science & Engineering Department,
University of Bridgeport, CT-06604, USA; phone: 203- 576-4703; fax: 203-576-4766;
that causes the energy damage . Alternatively, scheduled
based protocols use time division multiple access (TDMA)
mechanism to decrease the energy waste. From other side,
scheduled based MAC protocols experience the problem due to
Scalability and mobility of the modes . As a result, broken
Few cross layer MAC protocols are found in the literature,
which reduce the energy consumption by adjusting the reliable
link and bandwidth constraints . However, these protocols
experience co-channel interference due to long state transitions
. End-to-end delay can be minimized by combining the MAC
and network layer features . Also, end-to-end delay can be
guaranteed while choosing the best least delayed path .
The reported MAC protocols in the literature are not fully
capable of supporting mobility and network adaptability. These
are some of the major issues which need to be addressed when
designing a highly robust MAC protocol. To address these
concerns, PT model is integrated BN-MAC to provide the
mobility support . With integration of PT with BN-MAC,
the several WSN applications can be supported using less
energy consumption such as disaster, surveillance, monitoring,
home automation devices and controlling the remote devices.
In this paper, PT routing model provides cross layering support
for BN-MAC to handle the mobility and scalability over WSNs.
II. SYSTEM MODEL FOR BN-MAC PROTOCOL
BN-MAC protocol is an energy efficient that reduces the
energy consumption while handling idle listening, overhearing
and congestion. It also shortens the latency while guarantying
high reliability in a mobile environment .
Let us assume that system model should be composed of
many small nodes, which are organized in ad-hoc fashion. The
nodes should use short range and one-hop communication
rather than long range communication to save the energy. In our
case, we use 1-hop destination search for scheduling and
sending the data. The WSN in system model is divided into
different regions, and each region is controlled by a boarder
node (BN) as shown in Figure 1. The BN plays a role as a
coordinator to forward the data within the region and the
Pheromone Termite (PT) Model to provide
Robust Routing over Wireless Sensor Networks
1Abdul Razaque, Member IEEE and 2Khaled Elleithy, Senior Member IEEE
The message forwarding process of BN-MAC protocol
involves two types of communications: intra and inter. Intra
communication process is carried within the region while inter
is performed out of the region. The mode of communication
within the region is based on Anycast communication. The
Anycast communication reduces the latency as compared with
multicast communication. The multicast consumes more
energy while forwarding the packets, but larger packet size
severely affects the network performance during the
We preferred to use Anycast to reduce overhead of packet
forwarding from each node. The most of latest WSNs
applications are in the surveillance and monitoring area. For
such applications mobility and packet generation rate of the
network are mandatory constraints. If most of the nodes remain
in an idle state for a longer time, considerable amount of energy
is wasted. In our case, sensor node does not remain in idle state
because node finishes its monitoring process then goes to sleep
state using Automatic active and sleep model explained in .
We use BN-MAC with PT for controlling the static and
mobility devices from remote places depicted in Figure 1.
Figure 1. Proposed simulated scenario for handling the devices from a
remote distance over wireless sensor network (WSN)
III. PHEROMONE TERMITE MOBILITY-AWARE
MODEL FOR BN-MAC
As, we discussed earlier that BN-MAC protocol leverages
the features of carrier sense multiple access (CSMA) and
TDMA. The CSMA is based on semi synchronous mechanism
supported with low duty cycles. From other side, scheduled
based part uses the PT that provides the cross layering support
for finding the best route to forward and receive the data packets
at one-hop neighbor nodes.
Once the carrier medium is accessed, the sensor nodes fix
to schedule for sending the data based on using pheromone
mobility aware route. Let us assume that ‘Pl’ is the variable
length of packets forwarded to other neighbor nodes. The
distance between the two nodes is ‘r’. Thus, according to
Newton’s law of gravitation, the distance is inversely
proportional to force the . Therefore, we can apply free
space propagation model to measure distance between two
neighbor nodes based on the following parameters.
Dte: Default transmitted energy; Et: Energy gain of the
transmitter (TX); Er: Energy gain of the receiver (RX)
Lt: location of transmitter (Tx); Lr = location of receiver (RX);
pl: Received packet length; LN :Loss in network.
The calculated distance is used for updating the trajectory
pheromone of sensor nodes. We hereby deploy the features of
trail pheromone and ant control algorithm.
is the number of pheromones that source sensor
node‘s’ spreads on the link at the one-hop neighbor ‘l’ for node
‘n’. ‘Hp’ is the previous destined hop of packet, ‘Pa’ is the
amount of pheromone used in each destined packet. ’rc’ current
distance of neighbor node ‘n’ at link ‘l’ and ‘e’ is the distance
of same neighbor node ‘l’ when last packet received and ‘β’ is
the packet-generation rate. The calculated trail pheromone is
used to determine the forwarding energy power of each
neighbor node. Packet forwarding power of each neighbor node
can be calculated as follows:
Where, P(N)q,r is the energy power of each neighbor node ‘u’ to
forward the packet destination ‘r’ at node ‘n’ and ‘K’ is total
number of neighbor nodes. ‘C’ is a pheromone threshold that is
constant. Ps is the level of pheromone sensitivity. Pheromone
threshold and pheromone sensitivity can also be used to find the
second best alternate path of forwarding the packets to the
We here determine an average predictable amount of
pheromone ‘Pψ’ using different links. Let us assume ‘A’ is the
source node and ‘B’ is the destination node, which are using
two different links: for sending pheromone.
Each link consists of different attributes that are characterized
by non-negative random operation ‘λo(r)’ with mean value
Each packet the forwards fixed amount of pheromone ‘Pa’.
Let us assume that each node generates pheromone at constant
rate ‘β’. Suppose two nodes: ‘A’ and ‘B’ are located at two
different locations with distance ‘r’ which are uniformly
distributed over the network. Thus, Rayleigh Distribution can
be used to find the distance distribution of nodes. If
transmission power of the sensor node is less than WSN area,
then the distribution distance is divided into a range of the 0 to
r that can be calculated as:
This is the probability density function that is used to determine
the density of the WSN .
Where, ‘V(r)’ is the node distribution that can be used to
compute the predicted pheromone generation
between the node distribution distance ‘r’ with
respect to the number of arrived packets.
Let us assume ‘Z’ is random variable that is used to describe the
fraction of generation pheromone between node
distribution distance ‘r’ corresponding to packet arrival rate.
Thus, predicted amount of pheromone can be computed using
For 0 ≤ r ≤ R as follows:
Enumeration the order of equalities:
Thus, degenerated predictable pheromone can be calculated as
The predicted generation rate can be used to compute the
average predictable pheromone amount on a single and multiple
links using pheromone update-degeneration function. Let us
assume ‘P’ is the population at a distance ‘h’ and ‘Pi’ is the
initial population. Thus, the ‘P’ can be derived as follows:
The updated pheromone function can also be written as:
This function is used for calculating an average predicted the
pheromone amount on a single and multiple links. Based on the
following assumptions; an average predicted pheromone on the
single link can be determined using pheromone update equation
for the number of ‘n’ packets.
Number of delivered packets for distance’ r’ is Poisson
distribution with an average wavelength
a. The average amount of received pheromone is ‘ω’.
b. Initial pheromone amount ‘Pi’ on a single link.
Thus, the pheromone update equation is used consecutive times
Thus, the predicted pheromone amount on the single link for
node distribution distance ‘r’ for number of ‘n’ arrived packets
PP(r) is expressed with Poisson distribution amount
given as (11):
Where, λ: Average number of successfully received packets, Z:
Number of successful attempts. We map and apply the Poisson
distribution ‘ψ’ in our problem, and the details are given as
An average pheromone performance for a longer time can be
obtained as follows:
If we use only single link for destined the packets, then ‘
is the predicted pheromone amount on a single link. Let us
assume that the forwarded packets on multiple links as depicted
in Figure 2, showing the behavior of a termite when attempting
to find the food. Similarly, the PT works for WSNs for
providing the links on the path. Suppose, P0, P1, P2, ..,Pn be the
multiple links to forward the data over WSN. The amount of
predicted packet degeneration is .
Thus, when a packet is received by node ‘n’, it forwards the
packet to 1-hop neighbor nodes, and pheromone is degenerated
on all the links based on predicted packet degeneration rate.
Thus, the average pheromone for all the multiple links can
calculate as follows:
Figure 2. Single and multiple links to forward the packets using pheromone-
IV. SIMULATION SETUP AND ANALYSIS OF
We set up simulation scenario for controlling remote devices
over WSN. We use ns-2.35-RC7 that produce results that are
almost similar to real environments. In the experiments, the
WSN is divided into ‘N’ regions to get the information more
We have combined mobility- and static-based scenarios. The
main objectives of the simulation are to determine suitable
routing protocol for BN-MAC. Thus PT, EAP and SPIN are
integrated and simulated with BN-MAC. The simulation
scenarios consist of 300 nodes, which are randomly placed in a
geographical area of 600 × 600 m2. The area is divided into ‘N’
number of 150 × 150 m2 regions. The initial energy of each
sensor node is set to 20 J.
The bandwidth of the nodes is 40 kb/sec, and the maximum
power consumption for each sensor node is set 14 mW. The
sensing capability is 13 mW. Each sensor is capable of
broadcasting the data at a power intensity ranging from -16 to
13 dBm. The size of the packet is fixed to128 bytes. The sink
location in each region is at the distance of (45, 45). The node
mobility is set from 0 m/sec to 18 m/sec. The transmission range
of node is 30 meters with 10 meter sensing capability.
The total simulation time is 20 minutes, and the pause time is
set to 2 sec before start of the simulation. During this phase,
nodes are in warm up phase. The results are an average of 12
A. Control Packets
The routers consume a substantial amount of energy to send
control packets in WSN applications. The control packets do
not send any data except the handshaking process but consume
network bandwidth. An energy-efficient routing protocol can
minimize the number of control packets that are sent to save
energy and bandwidth. Figure 3 presents the control packet
overhead of PT, SPIN, and EAP with BN-MAC.
The number of control packets is directly proportional to
node mobility. PT outperforms SPIN and EAP. PT is a bio-
inspired protocol that does not vary under different mobility
conditions, whereas the other mobility protocols experience
problems due to changes of mobility. Furthermore, EAP and
SPIN suffer due to frequent link break-up because of high
mobility and thus require more control packets to re-establish
0 2 4 6 8 10 12 14 16 18
Figure 3. Control packets for PT, SPIN, and EAP for different
numbers of nodes
We evaluate the throughput performance of each routing
protocol. PT appears to be compatible with BN-MAC. Figure 4
presents the results of simulations using EAP, SPIN, and PT
with BN-MAC. To check the robustness of three routing
protocols, we simulate a scenario that involves static and
mobile objects. The speeds of the sensor nodes vary from 0 to
The simulations validate that BN-MAC with PT produces a
stable throughput, whereas SPIN and EAP with BN-MAC face
the slight problems due to motion. As a result, SPIN and EAP
have reduced the throughputs. The simulation results
demonstrate that PT with BN-MAC is the superior choice for
several WSN applications. BN-MAC-EAP and BN-MAC-
SPIN result in reduced throughput because both lack mobility
features and consume additional time during route discovery
and while maintaining the links.
0 2 4 6 8 10 12 14 16 18
Figure.4. Throughput of BN-MAC-PT, BN-MAC-SPIN and BN-MAC-EAP
on different mobility rates
In this paper, a scalable and a mobility-aware pheromone
termite (PT) model is presented to provide robust and faster
routing over WSNs. The model supports single and multiple
paths over WSNs. Two important features: packet generation
rate and pheromone sensitivity are analytically discussed. BN-
MAC-PT is compared with BN-MAC-EAP and BN-MAC-
SPIN using ns2 simulator to analyze the strength of PT
The simulation results demonstrate that BN-MAC-PT is the
superior choice for mobility and scalability where it achieves
15-20% higher throughput at different mobility rates. In
addition, PT-BN-MAC sends 22-27% fewer control packets as
compared with other routing protocols. As a result, each node
saves 13-18% energy.
 A. Razaque and K,M. Elleithy, “Automatic energy saving (AES) model
to boost ubiquitous wireless sensor networks (WSNs), International
Journal of computers and technology (IJCT),vol.10, no.5 ,pp. 1640-1645,
 A. Razaque and K, M. Elleithy, “Energy-Efficient Boarder Node Medium
Access Control Protocol for Wireless Sensor Networks”, journal Sensors,
Vol-14(3), March, 2014.
 J. Lutz et al.,” Apples and oranges: comparing schedule- and contention-
based medium access control” In proceedings of the 13th ACM
international conference on Modeling, analysis, and simulation of
wireless and mobile systems (MSWIM '10), pp. 319-326, 2010.
 Pei. H, Li. X, Soroor.S, Matt. W. M and Ning.X,” The Evolution of MAC
Protocols in Wireless Sensor Networks: A Survey”, IEEE Trans. on
communication surveys and tutorials, Vol.15, no.01, pp. 101-120, 2013.
 .S.Liqi and O.F.Abraham, “.TDMA Scheduling with Optimized Energy
Efficiency and Minimum Delay in Clustered Wireless Sensor Networks”,
IEEE Trans. Mobile Computing, Vol. 9, No. 7, pp. 927-939, July 2009.
 Liqi Shi, and Abraham O. Fapojuwo, ―TDMA Scheduling with
Optimized Energy Efficiency and Minimum Delay in Clustered Wireless
Sensor Networks‖, IEEE Transaction on Mobile Computing, Vol. 9, No.
7, pp. 927-939, July 2009.
 K Sumit, C Siddhartha, “A Survey on Scheduling Algorithms for Wireless
Sensor Networks”, International Journal of Computer Applications,
Vol.20, N0.5, 2011.
 Zara. H and Faisal.B, “XL-WMSN: cross-layer quality of service protocol
for wireless multimedia sensor networks”, EURASIP Journal on Wireless
Communications and Networking, 2013.
 E Felemban, C-G Lee, E Ekici, MMSPEED: multipath multi-SPEED
protocol for QoS guarantee of reliability and timeliness in wireless sensor
networks. Mobile Comput. IEEE Trans 5(6), 738–754 (2006).
 A. Razaque and K. Elleithy, “Least Distance Smart neighboring Search
(LDSNS) over Wireless Sensor Networks” , In proceeding of IEEE
international conference on European Modelling Symposium EMS2013,
Manchester, United Kingdom, 20 – 22 November 2013.
 Ming. L.I et al., “An Energy-Aware Routing Protocol in Wireless Sensor
Networks”, International Journal of sensors, pp.445-462, 2009.
 A. Razaque and K M. Elleithy “Mobility-Aware Hybrid Medium Access
Control Protocol for Wireless Sensor Network (WSN)”, In proceedings
of IEEE international conference on Sensors Applications Symposium
(SAS), Rydges Lakeland Resort, New Zealand, 18-20 February, 2014.
 A. D. Taskok, “A derivation of Newton’s law of gravitation from
electromagnetic forces”, Bulg. J. Phys. 30, pp. 7–20, 2003.
 Wolfram math world from: /http://mathworld. wolfram.com /
Mr. Abdul Razaque is a Phd student of computer
science and Engineering department in the
University of Bridgeport. Mr. Razaque has research
interests in the development of mobile applications
to support mobile collaborative learning (MCL),
congestion mechanism of transmission of control
protocol including various existing variants, and
delivery of multimedia applications. He has
published over 60 research contributions in refereed
conferences, international journals and books.
He presented his work in more than 30 countries. During the last two years he
has been working as a program committee member in IEEE, IET, ICCAIE,
ICOS, ISIEA and Mosharka International conference. Abdul Razaque is
member of the IEEE and ACM.
Dr. Khaled Elleithy is the Associate Dean for
Graduate Studies in the School of Engineering at
the University of Bridgeport. He has research
interests are in the areas of network security,
mobile communications, and formal approaches
for design and verification. He has published more
than two hundred fifty research papers in
international journals and conferences in his areas
Dr. Elleithy is the co-chair of the International Joint Conferences on Computer,
Information, and Systems Sciences, and Engineering (CISSE). CISSE is the
first Engineering/Computing and Systems Research E-Conference in the world
to be completely conducted online in real-time via the internet and was
successfully running for six years. Dr. Elleithy is the editor or co-editor of 12
books published by Springer for advances on Innovations and Advanced
Techniques in Systems, Computing Sciences and Software.