Multiagent based multipath routing in wireless sensor networks

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 6, No. 2, April 2014
DOI : 10.5121/ijwmn.2014.6207 67
MULTIAGENT BASED MULTIPATH
ROUTING IN WIRELESS SENSOR
NETWORKS
Vidya S. Bennur1
, Sadashiv S. Shirabur2
and Ashok V. Sutagundar3
1
Department of Electronics and Communication Engineering, East Point College of
Engineering for Women, Bengaluru, Karnataka, India - 560049
2
Department of Electronics and Communication Engineering, Veerappa Nisty
Engineering College, Shorapur, Karnataka, India - 585220 3
Department
of Electronics and Communication Engineering, Basaveshwar Engineering College
Bagalkot, Karnataka, India – 587102
ABSTRACT
This paper proposes a Multiagent Based Multipath Routing (MBMR) using a set of static and mobile agents
by employing localization technique. The operation of proposed routing technique can be briefly explained
as follows. (1) Anchor nodes are deployed evenly over the network environment. (2) Unknown sensor nodes
are deployed randomly over network environment and these nodes perform localization. (3) Source node
computes the shortest route to destination node through arbitrary midpoint node and intermediate nodes.
(4) Source node generates mobile agents for partial route discovery, which traverses to destination node
through the midpoint and intermediate nodes by carrying information. (5) Mobile agents update the
destination node with carried information. (6) Destination node computes route weight factor for all the
routes discovered by mobile agents. (7) Destination node computes the node disjoint routes and it selects
routes depending on the criticalness of event for communication. The performance of the proposed scheme
is evaluated in terms of performance parameters such as localization error, network lifetime, energy
consumption, cost factor, packet delivery ratio, and latency.
KEYWORDS
Wireless Sensor Networks, Localization, Agents, Multipath Routing
1. INTRODUCTION
In the recent years, the advances in VLSI technology, Micro-Electro-Mechanical (MEMS) system and
low power radio technologies have created low power, low cost and multifunctional wireless sensor
devices, which can observe and react to changes in physical phenomenon of their environment. WSNs
consist of a very large number of small, inexpensive, disposable, robust and low power sensor nodes
working co-operatively. The sensor node consist of sensing unit to sense the environment, processing
unit to process the sensed raw data, memory, battery and transceiver for communication [1-4]. These
sensor nodes are scattered across the geographical area with larger number nodes to sense over an area
so that they can provide greater accuracy. Typically the sensor nodes coordinate themselves to
perform a common task, and are capable of collaborating with one on another and measuring the
condition of their surrounding environment. The sensed measurements (data) are then transformed in
to digital signal and processes to reveal some properties of the phenomenon around the sensors. Due to
the fact that sensor nodes in WSNs have short radio transmission range, the intermediate nodes act as
relay nodes to transmit data towards the destination node (sink node). The sensors can be used to
monitor wide variety of ambient conditions (parameters) such as temperature, humidity, pressure,

68
salinity and mobility. This monitoring capacity of sensors can be effectively used in applications like
military (battlefield surveillance) application, environment (fire detection, earthquake monitoring)
application, agriculture irrigation monitoring, and health applications [5].
We briefly describe the different types of wireless networks such as cellular and ad-hoc networks in
order to differentiate the WSNs from them. The cellular network is one which consists of cellular
phones those are mobile devices and roaming over an area, the area is divided into smaller regions
called as cell and each cell is equipped with a Base Station (BS). The BS of all the cells is connected
to a backbone network. The mobile devices communicate each other by establishing a connection
through the BS. The cellular network uses packet switching mode operation in recent days, but
traditionally it uses circuit switching. The ad-hoc network is an infrastructureless network because the
sensor devices are deployed over an area without the infrastructure. All the devices are mobile and
communication is relayed on many mobile devices so each device can act as user and router for data
transmission. These networks are established in an emergent situation over a small geographical area.
However the mobile devices are rechargeable. Sensor networks consists of large number of nodes in
the network, deployed densely which can be several orders of magnitude higher than the nodes in an
ad-hoc network. Sensor nodes are more prone to failures. Topology of sensors network changes very
frequently compared to ad-hoc network. Sensor nodes mainly use broadcast communication paradigm
than point to point communication. Sensor nodes are limited in power, computation capabilities,
memory and most of the time batteries of the nodes are not rechargeable [6].
In traditional WSNs, the sensors nodes continuously at a periodic time gather the data and transmit it
to the destination node by multi-hop communication. The conventional method of data gathering and
processing in WSNs may lead to some of the limitations like increased latency, excess energy
consumption, bandwidth overhead and redundant data transmission etc. The important issue in WSNs
is the network lifetime which is mainly dependent on the energy of sensor nodes.
Recently, the multipath routing is widely used by many researchers to develop the best networking
protocols to improve the network performance by utilizing the network resources effectively. As the
radio transmission range of sensor node is short, the communication between the source and
destination is accomplished by multi-hop communication. But the communication cost is usually
several orders of magnitude higher than the computation cost. During routing phase, the sensed data,
query and control messages travel from a node to another node to reach the destination node. This
transfer consumes a considerable amount of sensor energy. Therefore many routing protocols are
proposed for conserving the sensor energy and to prolong the lifetime of network. But many of these
protocols use conventional method for the route discovery which also consumes a considerable
amount of sensor energy.
Agents have been used in many applications to reduce the number of message that need to be send
from one sensor to another. This paper proposes the routing protocol in WSNs using agents to reduce
the number of data and control messages and prolong the sensors lifetime. Agents are also great tools
in case of communication with low bandwidth which is one of the main features of sensor networks.
The WSNs are large scale networks with hundreds and even thousands of highly resource constrained
sensors nodes put together in network. For example in a network, sensors are densely distributed
across a large building or over an area such as forest, the typical tasks for such a network are, to send a
data to a sensor which may be a destination node, but this task requires the knowledge of position of
the sensor nodes or at least relative positions of sensor nodes. To know the position of each sensor
node it is possible to equip the each sensor node with a Global Position System (GPS) which provides
them with absolute position, but this is currently a costly solution. Because in WSNs the energy
conservation has been considered as a core issue and cost, as well as size of the sensor node should be
as small as possible. So these constrains of WSNs is not overcome just by equipping GPS to each
sensor node.

69
Unfortunately, for a large scale network with hundreds or thousands of sensor nodes the solution is not
adding GPS to all sensor nodes in the network and it’s not feasible because [7]. 1) GPS consumes
more energy for its operation which inturn reduce the life time of WSN. (2) In large scale WSN if
each sensor node is equipped with the GPS, then the production cost of the network will be very high.
(3) Size of sensor nodes are required to be small but the size of the GPS and its antenna increases the
sensor node size. (4) In the presence of dense forests, mountains or other obstacles that block the line-
of-sight from GPS satellites, GPS cannot be implemented.
For these reasons an alternate solution of GPS is required which is cost effective, energy conservable,
effective in sensor size and can operate in diverse environments. The multipath routing algorithm
proposed in this paper requires sensor nodes to be able to locate themselves in co-ordinate system in
various environments. This problem, to which we refer as localization, is a challenging issue in WSNs
to improve the quality of network in terms of energy consumption, size and cost [8][9][10]. The
localization opens up new ways of reducing power consumption in multi-hop WSNs. To perform the
localization, the proposed routing protocol uses the power of the received beacon signal of the three
anchor nodes and with the aid of trilateration method the position of node is determined.
Many of the subsisting research presented the routing protocol for WSNs which are dealing with:
energy awareness to maximize the lifetime of WSNs, fault tolerance in case of nonworking sensor
nodes and scalability for large number of sensor nodes. Some of the critical application such as fire
detection, disaster management and military application etc need energy constraint, delay constraint
routing, size of the sensor nodes as small as possible and network production cost should be as small
as possible.
1.1 Related Work
Some of the related works of localization technique are as follows: Estimating the geographic position
of sensor node is a crucial issue in WSNs to reduce the power consumption, size and cost of wireless
device without equipping more of sensors with GPS. The localization from mere connectivity presents
a method to determine the location of nodes in the network by using the connectivity information [11].
It uses multidimensional scaling to determine the location of nodes.
The work given in [12] describes a localization algorithm based on the angle of arrival of beacons
from three or more fixed beacon nodes. In this technique to reduce the hardware complexity it uses
direction estimation of received beacons. The beacon node location and these angular directions are
used by standard triangulation technique to determine location of sensor node. Ad-hoc positioning
system uses angle of arrival capability of nodes with triangulation technique to determine the
orientation and position of nodes in ad-hoc network [13].
A multi-hop localization technique for WSNs using acquired Received Signal Strength Indication
(RSSI) is presented in [14]. In this technique the RSSI packet is exchanged between the nodes and
based on this a ranging model is constructed then it is used by minimum least square algorithm to find
location of node. The work given in [15] describes a GPS less localization scheme for outdoor
environment based on connectivity metric. The connectivity metric uses the Radio Frequency (RF)
communication capabilities of wireless devices. A fixed number of reference points are used for
localization with overlapping coverage regions and transmit periodic beacon signals. Nodes use a
simple connectivity metric to localize themselves to nearest reference points.
The fully decentralized Anchor Free Localization (AFL) algorithm is presented in [16]. The nodes in
AFL start from a random initial co-ordinate assignment and converge to a consistent solution using
only local node interaction. The robust positioning algorithm for distributed ad-hoc WSNs is depicted
in [17]. The localization algorithm works in two phases, in first phase it uses hop-terrain algorithm
with triangulation to find the node position. In second, refinement phase the more accurate position of
node is computed by least square triangulation technique which is an iterative algorithm. A practical
localization algorithm for WSNs is presented in [18]. It works in two phases, first, it partition the

70
target region by dividing into small grids. Second, the refinement phase in which more accurate
location of node is determined by applying trick algorithm.
The work given in [19] describes an improvement to Distance vector (DV)-hop algorithm. It presents
a new DV-hop algorithm based on optimal nodes. A RSSI based DV-hop algorithm for WSNs is
depicted in [20]. For localization each node uses three or more beacon signals from anchor nodes. But
in this algorithm each sensor node uses hop count from two anchor nodes and RSSI of another anchor
node for location identification. The work in [21] describes the location tracking in indoor
environment using the RSSI of beacon nodes with trilateration estimation technique.
Some of the related works of multipath routing are as follows: Energy based multipath routing
protocol for WSNs describes how the network life time can be maximized through balancing energy
consumption across multiple sensor nodes [22]. It uses service differentiation concept to provide the
accepted delay for delay sensitive application. This protocol employs a queuing model to differentiate
between real and non-real time traffic. It reduces the end-to –end delay by distributing traffic across
multiple paths and increases the throughput rate to shortest path.
The work given in [23] presents the awareness of Quality of Service (QOS) and energy efficiency by
multipath routing which increases the network lifetime in WSNs. It establishes a tradeoff between
energy and QOS to improve the network performance. Relax multipath routing protocol for WSNs is
presented in [24]. This protocol uses battery relaxation period to improve the network lifetime and
reliability. The relaxation period is used to recover the battery from its lost capacity to increase the
battery life, which increase the lifetime of entire network. To provide reliable service, the source node
periodically floods the keep alive message over multiple paths to keep them alive.
The [25] describes the energy efficient and QOS based routing protocol for WSNs. It achieves load
balancing through splitting up the traffic across a set of available node disjoint paths. This protocol
increases the reliability of data delivery through utilization a light wait XOR based foreword error
correction technique to provide data redundancy. It uses the residual energy; buffer size and signal to
noise ratio to predict the next hop through the path construction phase.
The congestion control in WSNs based multipath routing in priority rate adjustment technique
protocol is presented in [26]. The congestion in WSNs needs to be controlled to prolong the lifetime of
network and to avoid packet loss. It is required to improve the fairness and provide better QOS. The
efficient multipath protocol for WSNs describes a multipath routing algorithm based on direct
diffusion that reinforces multiple paths with link quality and low latency [27]. It improves the energy
efficiency and scalability.
When certain nodes and links become over utilized causes congestion, the traffic can be spread over
the alternative paths to balance the load over those paths and increases the degree of fault tolerance.
The work given in [28] describes design of an alternate multipath routing protocol for WSNs when the
nodes on primary path fail. The work given in [29] depicts a design of a multipath energy efficient
congestion control scheme to reduce the packet loss due to congestion. It presents accurate detection
of congestion and its control. For congestion estimation it uses combination three parameters queue
size, contention and traffic rate. To control the congestion it uses two techniques depending on degree
of congestion defined by three parameters, which are rate control technique and multiple energy
efficient paths.
To enable reliable data delivery a multipath routing algorithm is presented in [30]. The packet loss and
congestion in the network can be reduced by controlling the scheduling rate of packets depending on
priority of data. The fault diagnosis is a important issue in WSNs to reduce energy consumption by
faulty nodes and to ensure the gathering information accuracy. An energy efficient data collection
protocol which consists of clustering and multipath routing is described in [31]. The clustering based
on fault diagnosis avoids the faulty nodes becoming a cluster heads so that it reduce the energy
consumption and fault information transmission.

71
The work given in [32] describes that continuously using the minimum energy path deprives the nodes
energy and time taken to determine an alternate path increases. The energy efficient adaptive
multipath routing technique is proposed, the adaptive is because they have low routing overhead. This
protocol provides reliability and low energy consumption. Agent based multipath routing presented in
[33] which describe the use of agents to reduce the complexity in routing technique. The agents also
reduce the communication cost.
The work given in [34] describes the data gathering and aggregation to avoid the redundant data
transmission and to reduce the bandwidth required for data transmission. To perform aggregation of
data it constructs a wheel with spokes around the event triggered node. The energy consumption
reduced by using the agents for path discovery. It provides reliability for data transmission by multiple
paths.
Some of the drawbacks of existing localization technique are as follows: lack of mechanisms for
anchor node deployment, time of arrival based localization technique needs extra hardware otherwise
a small error in time causes larger error in distance estimation and angle of arrival based localization
technique needs extra directional antenna and it is not advisable for large WSNs. And drawbacks of
existing routing protocols are: use of GPS to all nodes increases the size of wireless device, cost and
energy consumption which leads to reduced network lifetime, conventional method of route discovery
and route maintenance consumes more energy, lack of robust mechanism for critical information
transmission, lack of mechanism for selecting good relay nodes for route construction.
We have proposed a MBMR in WSNs by employing localization. The localization technique works
based on the received power of beacon packet from anchor nodes. These beacons are prioritized based
on their power level and three highest power beacons are used to distance estimation with appropriate
propagation model. Due to use of localization scheme the size of the some nodes is reduced and cost
of network deployment is also reduced. As the GPS is used only by anchor nodes the energy
consumption by GPS in unknown nodes is reduced.
1.2 Our Contribution
This paper proposes MBMR scheme in WSNs. The proposed scheme uses two types of software
agencies which are sensor and destination node agencies and those are static and mobile.
The operation of the proposed routing scheme can be explained as follows: (1) Anchor nodes are
deployed evenly over the entire network. (2) Unknown sensor nodes are randomly deployed over the
entire network environment. (3) Now each anchor node broadcasts a beacon signal by flooding (up to
2 hops). (4) Unknown sensor nodes perform localization to know its location co-ordinates. (5)
Whenever a sensor node detects an event, than it announces that, it is the source (event) node to its
neighbors and it computes the shortest distance to destination node using GPS location information
and its location co-ordinates. (6) Source node computes the arbitrary midpoint node between source
node and destination node using Euclidean distance. It also determines the intermediate nodes above
and below the midpoint node and also calculates the ascending and descending angles. (7) Source
node generates mobile agent and its child; each one will move on midpoint node and intermediate
node above and below the midpoint node by carrying the information like node id, residual energy,
hop count and hop distance, event information etc. (8) The mobile agent and its child which are
traversed through the network by carrying the route and network information (node id, residual
energy, hop count, available bandwidth, hop distance and event information) will updates the
destination node with this information. (9) Destination node computes the path weight factor based on
hop distance factor, link efficiency and energy efficiency for all the routes discovered by the mobile
agents and assign to the respective routes. These routes are prioritized based on their route weight
factor and it identifies the node disjoint routes. (10) Destination node evaluates the criticalness of
event based on the information carried by the mobile agents. Depending on the criticalness of event

72
the destination node selects a single route for in time data delivery or multiple routes for robust and
reliable communication. It also triggers mobile agent and its clones to establish the required routes.
Our contributions in the proposed routing technique are as follows: (1) Anchor node deployment
evenly over the network environment. (2) Computation of received power from anchor nodes at each
unknown sensor nodes. (3) Trilateration technique is applied based on received power at unknown
sensor nodes to know their position. (4) Method to select midpoint node and intermediate nodes. (5)
Usage of static and mobile agents for route discovery. (6) Route weight factor calculation based on
hop distance factor, link efficiency and energy efficiency. (7) Identification of node disjoint route. (8)
Best route/s selection based on event criticalness.
2. MULTIAGENT BASED MULTIPATH ROUTING
In this section we describe the preliminaries, network environment, models and agencies used for
proposed MBMR in WSNs.
2.1. Preliminaries
Some of the basic key terms used in proposed scheme are as follows.
a. Anchor node: It is a sensor node which has prior knowledge of its location co-ordinates
when it is deployed in the network environment since it is equipped with GPS.
b. Lateration: It is a technique, where the unknown sensor node location is determined through
distance measurement from unknown sensor node to anchor nodes.
c. Event: It is an action or occurrence detected by interaction of sensing unit with the
environment by sensing or controlling physical parameters and it is worth report to
destination (sink) node.
d. Event affected area: It is the area in which sensor node detects an event.
e. Source node: It is the sensor node in the event affected area which detects event first
compared to other nodes in that area. It announces itself as a source node to other nodes to
avoid the assuming the role of source node by others.
f. Mid point: It is the arbitrary middle point on the straight line joining (reference axis) the
source node and destination node.
g. Intermediate nodes: These are sensor nodes located above and below the (reference axis)
midpoint node between source and destination nodes and these intermediate nodes are all on
the same line with midpoint node.
h. Ascending angle: It is the angle formed between the midpoint node and intermediate node
above the reference axis at source node. The agents and messages moves with the ascending
angle from source node to intermediate node and with descending angle from intermediate
node to reach the destination node.
i. Descending angle: It is the angle formed between the midpoint node and intermediate node
below the reference axis. The agents and messages moves with the descending angle from
source node to intermediate node and with ascending angle from intermediate node to reach
the destination node.
j. Hop count: It is the total number of relaying nodes between the source node and destination
node.
2.2. Network Environment
The network environment is shown in figure (1), which consists of tiny sensor nodes with
heterogeneous sensing capability and destination node. In this network environment the sensor nodes
are classified in two types which are unknown sensor node and anchor node. The unknown sensor
nodes comprises of processor for preprocessing the sensed data, transreceiver for transmission and
reception of data and memory. These nodes are deployed randomly over the entire network
environment. The anchor nodes comprises of all the components of the unknown sensor node, in

73
addition to that it is having a GPS for location awareness. These nodes are deployed evenly over the
entire network environment. Immediately after the sensor nodes deployment the localization is
performed over the entire network to know the location of unknown sensor nodes by itself. After
localization, whenever a sensor node detects an event than it sends data to the destination node by
multi-hop communication. We assume that all sensor nodes are stationery for their lifetime and they
have same energy during the deployment phase. All the sensor nodes have same transmission range. It
is assumed that sensor nodes have capability to reconfigure the transmission power. Each sensor node
consists of an agent platform with security feature to coordinate the inter-agent communication.
Figure 1: Network Architecture
2.3. Models
This section presents the models used for proposed scheme such as anchor node deployment,
localization scenario, event detection, path parameters, and partial route discovery and route
establishment.
2.3.1. Anchor Node Deployment
We consider a squared area as a network environment for wireless sensor network
where . In this squared area identical unknown sensor nodes are deployed randomly, but
the anchor nodes are deployed evenly over the entire network environment. For performing the
localization each unknown sensor node should receive beacon signals from minimum three anchor
nodes. Due to this reason the anchor nodes should be deployed at particular locations over the network
environment. The anchor nodes can be deployed based on the equation and .
Where is the length and is breadth of network area, where and gives the
deployment location co-ordinates of each anchor node by varying the value of and in above
equation. The anchor nodes can be deployed up till . The is a numerical value
which should be chosen such that it divides the network area uniformly depending on the number of
anchor nodes so that anchor nodes can be deployed evenly over the network. The network
environment after the deployment of anchor nodes is shown in figure (2).

74
Figure 2: Anchor Node Deployment
2.3.2. Localization Scenario
To perform the multipath routing by proposed scheme in WSNs each sensor node in the network
should know their location co-ordinates, so to determine the location co-ordinates of unknown nodes
the localization is performed, which is as follows. After the deployment of all sensor nodes, each
anchor node broadcasts a beacon signal by flooding (up to two hops). The unknown sensor nodes
receive the beacon signals from different anchors. We assume that each unknown sensor node receives
more than three beacon signal from different anchors. Now each unknown sensor node computes the
power received by the reception of every beacon signal using the equations and .
Where is the total received power of beacon, is the rms voltage of the received beacon signal,
is the peak voltage (maximum of absolute value of signal) of the received beacon signal and is
receiving antenna resistance. The beacon signals are prioritized based on their received power.
Now the three beacons which are received with maximum power are used to find the
distances to their respective anchors from the unknown sensor node. According to empirical
propagation measurements in a radio channel shows that the average received signal strength at any
point decays as a power law of the distance of separation between a transmitter and receiver. So the
distance between the transmitter and receiver can be calculated by using the equation given by .
Where is power received at distance , is close in reference distance from the transmitter,
is path loss exponent.
We assume that the close in reference point = 1m and can be generally determined by
using free space propagation model given by FRIIS transmission formula given by equation .
Where is transmitted power, is transmitter antenna gain, is receiving antenna gain, is
system loss factor not related to propagation of data , assume (indicates no loss in
system hardware), is wavelength in meters .

75
After estimating the distance to three anchor nodes the trilateration estimation is applied to find the
location of unknown sensor nodes. The localization scenario is shown in figure (3) for one unknown
sensor node (U8) and same will be performed by each unknown sensor node.
Figure 3: Localization Scenario
Localization Algorithm
Nomenclature: Pr : Total power of a received beacon, N : Number of beacons received by the
unknown node, temp : Temporary variable, i and j : Index values.
1: Begin
2: Anchor node broadcast beacon signal;
3: Unknown node receives beacon signals;
4: for i=0 to N do
5: Calculate Pr;
6: Pr[i] = Pr;
7: end for;
8: for i = 0 to N do
9: for j =0 to N do
10: if (Pr[i] < Pr[j]) then
11: temp = Pr[i];
12: Pr[i] = Pr[j];
13: Pr[j] = temp;
14: else
15: go to step 9;
16: end if;
17: end for;
18: end for;
19: Use trilateration algorithm;
20: end;
2.3.3. Trilateration
The lateration is a most common method for deriving the position co-ordinates of unknown nodes,
where the trilateration uses the distance measurements to minimum three anchor nodes. This
estimation considers these distances between the reference location (anchor) and unknown location as
the radii of circles with centers at every respective reference location (anchor location). Thus the
unknown location is the intersection of the three sphere surface as shown in figure (4).
As shown in figure (4) the unknown node is located using the co-ordinate of
anchor nodes , , and distances between these nodes and

76
unknown sensor node , , . By using the above information we can derive the following system
of equations.
In practice, usually above set of equations has no solution as they are nonlinear. The above set
of equations can be converted to linear form of equation by subtracting last equation from first
two equations and .
Re-ordering the terms in equation and gives a proper system of linear equation as below.
Now the above set of equations and can be represent in the linear matrix form as
Where
And
According to equation , the co-ordinate of unknown sensor node can be obtained by
using estimation method of standard minimum mean square estimation, which is given by
equation .
Figure 4: Trilateration
Trilateration Algorithm
Nomenclature: Nu : Number of unknown nodes, (xi, yi) : Co-ordinates of unknown nodes, (xj, yj) :
Co-ordinates of anchor nodes, dj : Distance between unknown node and respective anchor node, X :
Matrix of co-ordinates of unknown node, i and j : Index values.
1: Begin
2: for i = 0 to Nu do
3: j = 0;
4: Calculate dj;
5: (xi - xj)2
+ (yi - yj)2
= dj
2
;

77
6: j = j+1;
7: Calculate dj;
8: (xi - xj)2
+ (yi - yj)2
= dj2;
9: j = j+1;
10: Calculate dj;
11: (xi - xj)2
+ (yi - yj)2
= dj
2
;
12: Subtract step 11 equation from step 5 equation;
11: Subtract step 11 equation from step 8 equation;
14: Rearrange the equations of step 12 and step 13;
15: Step 14 equation can represented in the form AX=B;
16: Apply standard minimum mean square estimation X = (AT
A)-1
AT
B;
17: end for;
18: end;
2.3.4 Event Detection
The logical relationship between the various parameters sensed by sensor node decides the occurrence
of an event. The event detection is performed as follows. Let , be the
event with associated sensed parameter values and be the
sensor parameter threshold values for event , where and
The sensed parameter values for the event are either set to 1 or 0 based on
event threshold as given by equation [34].
An occurrence of an event is given by equation .
3. ROUTE PARAMETERS
This section depicts the computation of route parameters such as link and route efficiency, energy
ratio and hop distance factor used in the proposed scheme.
Link efficiency can be computed as follows. Let be the discrete time discrete valued
channel capacity, be the bit rate of the channel, Ei be the total energy consumed for transmission
of a bit in link and be the signal to noise ratio [35]. Capacity of channel is given by the
equation .
Let be the energy consumed for the transmission of a bit per distance and can be computed by
using the following equation [35].
Link efficiency LK can be determined by using the equation [36].
Route efficiency can be determined by using the equation by considering links on the route
[37].
Energy ratio in each of the route from source node to destination node is computed using the equation
and [37].

78
Where to R be the residual energy of the intermediate nodes of a route. The minimum
and maximum are the residual energy for nodes in the route computed using above
equations . Now find route energy efficiency by using the equation .
The hop distance factor of a route is determined by using the equation .
Where is the Euclidean distance between source and destination node which is given by the
equation and is the total number of hops in route from source node to destination node.
3.1 Partial Route Discovery
Partial route discovery is performed by mobile agent (RDA) between the source node and destination
node U40(S) and D respectively. The source node determines the arbitrary midpoint node
between source node and destination node with the aid of its locations
and respectively on the reference axis by using the equation and and
node U24 also lies on the reference axis as shown in figure (7) [37].
Source node finds intermediate nodes which are above and below the reference axis midpoint
node. Let and be the location of intermediate nodes
above and below the reference axis respectively. Where and
and , are computed using the equation and respectively.
Where and are degree of neighbour and communication range of sensor node respectively.
Source node determines the ascending and descending angle for movement of mobile agent by using
the equation and .
We consider and to be alternate angles as shown in figure (7).
Mobile agent (RDA) and its clone (replica of mobile agent) migrate for partial route
discovery from source node to destination node on three routes as shown in figure (7). Mobile agents
while migrating from one sensor node to next sensor node it collects the residual energy of node,
distance between the nodes, hop count and deliver it to the destination node. Destination node uses
this information to compute route weight factor and node disjoint routes.
3.2 Establishment of Routes
To find the best routes between the destination node and source node, the destination node computes
node disjoint routes and route weight factor. It also determines the node disjoint routes by using the
partial route discovered topology information. Two routes are said to be node disjoint, if and only if
there is no common intermediate node between routes from source node to destination node.
Destination node uses dijkstra algorithm to find the route based on the shortest distance. The nodes in
the found route are deleted in the partial route discovered topology to find next route. The route

79
finding continues until there is no disjoint route. A route weight factor is assigned to each route which
is determined by using the equation .
Where . Depending on the priority of the parameters the value of can be chosen by
the administrator. For example if energy efficiency is given more importance than value will be
higher.
4. AGENCY
The proposed work uses a set of static and mobile agents for route discovery and route computation,
which are part of sensor node and destination node agencies. These are described as below.
4.1 Sensor Node Agency
The sensor node agency consists of Sensor Node Database (SNDB) for inter agent communication,
Sensor Node Administration Agent (SNAA) and Route Discovery Agent (RDA). The interaction
between the destination node agency components and its agents are depicted in figure (5).
Sensor Node Database: It resides in all the sensor nodes and it can be read and updated by the agents.
It comprises of node id, distance between the neighboring nodes, residual energy, active/sleeping
mode, destination node id, location information of destination node, upper and lower intermediate
node location information, signal strength, available bandwidth, event sensing time, distance between
itself and destination node, ascending and descending angles of intermediate nodes. The source node
maintains the route/s information from itself to destination node as given by the destination node
agency.
Sensor Node Administration Agent: It is a static agent resides in all sensor nodes of WSNs. It
maintains the information available such as signal strength, residual energy, transmission range etc. It
checks for event detection, if event is detected than it computes the arbitrary midpoint between the
source node and destination node and intermediate nodes above and below the reference axis between
source and destination node. It also computes the ascending and descending angle for agent
movement. The SNAA also computes the route parameters like link efficiency, minimum mad
maximum residual energy and hop distance factor. SNAA creates RDA and SNDB where it updates
all the route information and node information.
Route Discovery Agent: It is a mobile agent resides in all the sensor nodes. Whenever a node detects
the event than its SNAA generates RDA and its clone for route discovery. These RDAs migrates
through the midpoint node which is on the reference axis, intermediate nodes above and below the
reference axis with the help of ascending and descending angle to reach the destination node. These
RDAs while migrating through the intermediate nodes they carry the information of relay nodes such
as node id, residual energy, hop count, available bandwidth and hop distance; it also collects the
neighbour information. After reaching destination node RDAs updates the DNDB.
Figure 5: Sensor Node Agency

80
4.2 Destination Node Agency
The destination node agency comprises of Destination Node Database (DNDB) for interagent
communication, Destination Node Administration Agent (DNAA) and Route Setup Agent (RSA). The
interaction between the destination agency components and its agents are depicted in figure (6).
Destination Node Database: It is the database which resides in the destination node and it can be read
and updated by DNAA, RSA and RDA. It stores the network related information carried by RDA such
as node id, residual energy, hop count, hop distance, minimum and maximum residual energy of each
route, available bandwidth, channel capacity and DNAA information such as route weight factors,
criticalness of event and number of available routes.
Destination Node Administration Agent: It is a static agent which resides in the destination nodes. It
monitors the network related information such as number of active nodes, available route information
to reach source node. It computes the route weight factors based on the collected information from
RDAs such as route efficiency, route energy efficiency, hop distance factor and it also identify the
node disjoint routes between destination node and source node than it priorities the routes based on
their weight factor. Finally evaluates the criticalness of event, based on this it selects a single or
multiple routes to source node and triggers the RSA to establish the routes to source node.
Route Setup Agent: It is mobile agent which resides in all the sink nodes. It is triggered by DNAA
and it moves to source node by getting the required information from DNDB. After reaching the
source node it updates the sensor node database by route information.
Figure 6: Destination Node Agency
5. EXAMPLE SCENARIO
This section describes the operation of proposed scheme. It comprises of anchor nodes (A1-A9),
unknown nodes (U1-U81), source node (U40), destination node (D) and some agents for route
establishment. The sequence operation of the scheme is as follows.
1. Whenever an event is detected at source node U40, it initiates the route discovery process to
destination node.
2. SNAA computes the shortest route to destination node (U40-U28-U24-D), midpoint node U28,
intermediate nodes above (U23) and below (U27) the reference axis and angles for agent
movement.
3. SNAA generates RDA and its three clones. One RDA clone traverses on midpoint node and
remaining two will traverses above and below intermediate nodes respectively by carrying the
network , route and event related information.
4. RDA clones updates the DNDB after reaching the destination node (D).
5. DNAA computes the route weight factor and identifies the node disjoint routes and priorities
them.

81
6. DNAA evaluates the criticalness of event and selects multiple routes for reliable and robust
communication as the event is noncritical.
7. DNAA triggers RSA; RSA generates three clones to establish the selected routes between
destination node and source node.
Figure 7: Example Scenario
6. SIMULATION
To test the performance of the proposed scheme, the scheme is simulated using C programming
language. The proposed scheme is compared with an existing basic DV-hop localization scheme and
agent based multipath routing scheme [33]. In this section we describe the simulation model,
performance parameters and result analysis.
6.1 Simulation Model
Network model: We consider the network environment area of square meters for WSN. A
network consists of static unknown sensor nodes which are distributed randomly over the network
environment and anchor nodes placed evenly over the network environment. The bandwidth of
single hop connected sensor nodes is .
Channel model: For the proposed scheme the S-MAC protocol is used for media access [38]. We
assumed that packet transmission is in discrete time. Each node receives all packets from sending node
during its receiving interval unless the sender node is in non-active state. We have assumed that
channel is error free for simplicity.
Propagation model: For proposed scheme we used the free space propagation model with
propagation constant β. The transmission range of each node in WSN is R for single hop distance. It is
assumed that at any given time, the value of transmitted energy per packet is Eu Joules for every node.
R is directly proportional to Eu of the node, that is, R= CEu, where constant of proportionality C
depends on the medium of communication, attributed to β.
6.2 Simulation Procedure
Simulation parameters considered for analysing the scheme is given in table (1). Simulation procedure
involves following steps.
1. Generate the network environment.
2. Apply the localization scheme.
3. Apply the proposed routing scheme.
4. Compute the performance parameters of localization and routing scheme.

82
Table 1: Simulation Inputs
6.3 Performance Parameters
The following performance parameters are evaluated for analyses of the proposed schemes.
a. Localization error: It is defined as the ratio of summation of difference between the
estimated and original position of unknown sensor nodes to the total number of nodes, which
is given by the equation (37)
(37)
b. Network lifetime: It is the total number of rounds taken by the sensor node to die in the
WSNs.
c. Cost factor: It is defined as the ratio of number of anchor nodes to the total number of nodes
in the WSNs.
d. Energy consumption:
Energy consumption for localization: It is total amount of energy consumed to determine
the location of all the unknown sensor nodes in WSNs.
Energy consumption for routing: It is total amount of energy consumed for data
transmission, route discovery and route setup.
e. Packet Delivery Ratio (PDR): It is the ratio of packet sent to packet received.
f. Latency: It is the time required to transfer a packet from source node to destination sensor
node.
g. Number of routes: It is the total number of node disjoint routes between the source node and
destination node.
7. RESULTS
This section presents the results obtained during simulation. We compare the results of proposed
scheme with existing basic DV-hop localization scheme and Agent Based Multipath Routing (ABMR)
scheme for WSNs.
Localization Error: Figure (8) shows the localization error with increasing number of nodes and
increasing number of anchor nodes. The proposed localization scheme provides the reduced
localization error compared to basic DV-hop localization algorithm. As the number of nodes increases,
each unknown node get a beacon packet with better received power which increases the localization
accuracy their by reducing the localization error. As the number of anchor nodes increases the
unknown node get increased number of beacon packets, so that the location accuracy of unknown
nodes increases.

83
Figure (9) describes the localization error versus number of nodes with uniform and random
placement of anchor nodes. The proposed algorithm uses uniform placement of anchor nodes which
reduces the localization error compared to random placement of anchor nodes. when the anchor nodes
are placed randomly then some of the unknown nodes may not get the required number of beacon
packets for localization which increases the localization error.
Figure 8: Localization Error Versus Number of Nodes
Figure 9: Localization Error Versus Number of Nodes with Uniform and Random Placement of Anchor
Nodes
Network Lifetime: Figure (10) presents network lifetime for increasing number of nodes with and
without GPS. As the number of nodes increases the hop distance for communication between the
nodes reduces so that network lifetime of nodes increases as seen from simulation results. But in GPS
equipped nodes GPS should continuously monitor the node location so it consumes a considerable
amount of energy. Thus the proposed scheme uses very less number of nodes with GPS and remaining
are without GPS to increase the network lifetime.

84
Figure 10: Network Lifetime Versus Number of Nodes
Cost Factor: Figure (11) shows cost factor with increasing percentage of anchor nodes. As the
number of anchor nodes increases the cost of the network increases. Therefore to reduce the cost of
the network the proposed scheme uses a very few number of anchor nodes. .
Figure 11: Cost Factor Number of Anchor Nodes (Percentage)
Energy Consumption: Figure (12) presents the energy consumption for localization in milliJoules
with increasing number of nodes and increasing number of anchor nodes. As the number of nodes
increases the energy consumption reduces because communication cost is reduced by smaller hop
distances. But the computational cost increases which is negligible with respect to communication
cost.
Figure (13) depicts the energy consumption for routing in milliJoules with increasing number of
transmitting nodes and increasing communication range. As the number of nodes increases the energy
consumption increases. However, the proposed MBMR performs better compared to ABMR. Energy
consumption is due to gathering of partial topology information, route computation, route information
transmission, and reception. The proposed MBMR scheme exhibits less energy consumption because
it uses partial topology, hop distance factor, and energy information (energy efficiency) gathering for
route computation. ABMR uses only topology information and hazard nodes for route computation.

85
Figure 12: Energy Cosumption Versus Number of Nodes
Figure 13: Energy Consumption Versus Number of Transmitting Nodes
Packet Delivery Ratio: Figure (14) depicts the packet delivery ratio with increasing number of
transmitting nodes and communication range. As the number of nodes increases the PDR decreases
and increase in communication range increases the PDR because it reduces the number of isolated
nodes in the network. As the number of nodes increases, the data generated increases, which leads to
lack of bandwidth therefore the PDR reduces. But the proposed MBMR scheme provides better PDR
as compared to ABMR because it increases the bandwidth by considering link efficiency and
establishing only required number of routes.
Latency: Figure (15) explains the latency for increasing number of transmitting nodes and increasing
number of communication range. The proposed scheme performs better compared to ABMR. The
latency is due to route information gathering, computation, transmission and reception. In ABMR the
full topology information is used for route computation but in proposed MBMR the partial topology
information is used to compute the node disjoint routes.
Number of Routes: Figure (16) depicts the number of node disjoint routes established for increasing
number of nodes and increasing communication range. As the number of nodes increases, the number
of node disjoint routes also increases and also increase in transmission range, increases the number of
node disjoint routes.

86
Figure 14: Packet Delivery Ratio Versus Number of Transmitting Nodes
Figure 15:. Latency Versus Number of Transmitting Nodes
Figure 16: Number of Routes Versus Number of Nodes
8. CONCLUSION AND FUTURE WORK
We proposed the localization and multiagent based multipath routing scheme in WSNs.
In the ﬁrst work, we proposed a localization scheme to reduce the energy consumption, size of sensor
node and cost of network. The location of unknown node is determined by trilateration technique, by

87
the reception of beacon packets from anchor nodes. The anchor nodes are placed uniformly throughout
the WSNs to reduce the localization error.
In the second work, we proposed multiagent based multipath routing scheme in WSNs by employing
localization technique, set of static and mobile agents. In the proposed scheme a source node triggers
the route discovery mechanism by determining the intermediate nodes between source and destination
node. The set of mobile agents are used to carry the partial topology information like node id, hop
distance, residual energy and hop count etc., from source node to destination node through
intermediate nodes by using location information. The partial topology information is used in
computing the node disjoint routes by destination node. For node disjoint routes, the destination node
computes the route weight factor based on the route efficiency, energy ratio and hop distance factor.
For critical data, routing scheme assures intime data delivery by multiple routes; if non-critical then a
single route with highest weight factor is established to assure reliable data transmission.
There is a good future for application of WSNs. Many of the WSNs applications make the layman job
comfortable. But for these applications to become everyday reality a number of technological
challenges need to be addressed. This section brieﬂy explains the future work to proposed schemes.
For the proposed localization scheme, we can minimize the location error by employing any of the
error minimization technique (soft computing technique). Also we consider the anchor nodes as
mobile to reduce the location error. For the proposed routing scheme, we can consider some other
geometry for better performance of routing scheme. We can also consider the multiple destination
nodes or mobile destination node to improve the routing performance.
REFERENCES
1) Akyildiz, Ian F., et al., (2002) "A survey on sensor networks", Communications Magazine, IEEE 40.8, pp102-114.
2) Khemapech, I., I. Duncan, and A. Miller, (2005) "A survey of wireless sensor networks technology", 6th Annual
Postgraduate Symposium on the Convergence of Telecommunications, Networking and Broadcasting.
3) Rahman, Kazi Chandrima, (2010) "A survey on Sensor Network", Journal of Computer and Information
Technology, pp76-87.
4) Aboelaze, Mokhtar, and Fadi Aloul, (2005) "Current and future trends in sensor networks: a survey", Wireless and
Optical Communications Networks, 2005. WOCN 2005. Second IFIP International Conference on. IEEE.
5) Akyildiz, Ian F., et al., (2002) "Wireless sensor networks: a survey" Computer networks 38.4, pp393-422.
6) Papageorgiou, Pavlos, (2003) "Literature survey on wireless sensor networks"
7) Pal, Amitangshu, (2010) "Localization algorithms in wireless sensor networks: Current approaches and future
challenges", Network Protocols and Algorithms 2.1, pp45-73.
8) Langendoen, Koen, and Niels Reijers, (2003) "Distributed localization in wireless sensor networks: a quantitative
comparison", Computer Networks 43.4, pp499-518.
9) Amundson, Isaac, and Xenofon Koutsoukos, (2009) "A survey on localization for mobile wireless sensor
networks", Mobile Entity Localization and Tracking in GPS-less Environnments, pp235-254.
10) Pandey, Santosh, and Prathima Agrawal, (2006) "A survey on localization techniques for wireless networks",
Journal of the Chinese Institute of Engineers 29.7, pp1125-1148.
11) Shang, Yi, et al., (2003) "Localization from mere connectivity", Proceedings of the 4th ACM international
symposium on Mobile ad hoc networking & computing.
12) Nasipuri, Asis, and Kai Li., (2002) "A directionality based location discovery scheme for wireless sensor
networks", Proceedings of the 1st ACM international workshop on Wireless sensor networks and applications.
13) Niculescu, Dragos, and Badri Nath, (2003) "Ad hoc positioning system (APS) using AOA", INFOCOM 2003.
Twenty-Second Annual Joint Conference of the IEEE Computer and Communications. IEEE Societies. Vol. 3.
14) Alippi, Cesare, and Giovanni Vanini, (2006) "A RSSI-based and calibrated centralized localization technique for
Wireless Sensor Networks", Pervasive Computing and Communications Workshops, 2006. PerCom Workshops
2006. Fourth Annual IEEE International Conference on.
15) Bulusu, Nirupama, John Heidemann, and Deborah Estrin, (2000) "GPS-less low-cost outdoor localization for very
small devices", Personal Communications, IEEE 7.5, pp28-34.
16) Priyantha, Nissanka B., et al., (2003) "Anchor-free distributed localization in sensor networks", Proceedings of the
1st international conference on Embedded networked sensor systems.
17) Rabaey, C. Savarese J., and Koen Langendoen, (2002) "Robust positioning algorithms for distributed ad-hoc
wireless sensor networks", USENIX technical annual conference.
18) Huang, Tao, et al., (2010) "A Practical Localization Algorithm Based on Wireless Sensor Networks", Green
Computing and Communications (GreenCom), 2010 IEEE/ACM Int'l Conference on & Int'l Conference on Cyber,
Physical and Social Computing (CPSCom).
19) Qian, Qingji, Xuanjing Shen, and Haipeng Chen, (2011) "An improved node localization algorithm based on DV-
Hop for wireless sensor networks", Computer Science and Information Systems 8.4, pp953-972.

88
20) Tian, Shuang, et al., (2007) "A RSSI-based DV-hop algorithm for wireless sensor networks", Wireless
Communications, Networking and Mobile Computing. WiCom 2007. International Conference on. IEEE.
21) Pu, Chuan-Chin, Chuan-Hsian Pu, and Hoon-Jae Lee, (2011) "Indoor location tracking using received signal
strength indicator", Emerging communications for wireless sensor networks, pp 229-256.
22) Kumar, Ram, N. Vinothraj, and G. Kiruthiga, (2012) "Energy Based Multipath Routing Protocols for Wireless
Sensor Networks", Journal of computer application, Vol-5.
23) Rasouli Heikalabad, Saeed, et al., (2011) "QEMPAR: QoS and Energy Aware Multi-Path Routing Algorithm for
Real-Time Applications in Wireless Sensor Networks", International journal of computer science issues, Vol-8.
24) Devisri, R., and R. J. Archana Devy, (2011) "Reliable and power relaxation multipath routing protocol for wireless
sensor networks", Proceedings of international conference on advancement in information technology.
25) Ben-Othman, Jalel, and Bashir Yahya, (2010) "Energy efficient and QoS based routing protocol for wireless
sensor networks", Journal of Parallel and Distributed Computing 70.8 pp849-857.
26) Malar, R. Then, (2010) "Congestion Control in Wireless Sensor Networks Based Multi-Path Routing In Priority
Rate Adjustment Technique", proceedings of International Journal of Advanced Engineering & Applications,
pp28-33.
27) Li, Shuang, et al., (2010) "Efficient multi-path protocol for wireless sensor networks", International Journal of
Wireless and Mobile Networks 2.1, pp110-130.
28) Pratheema, S., K. G. Srinivasagan, and J. Naskath, (2011) "Minimizing end-to-end delay using multipath routing
in wireless sensor networks", International Journal of Computer Applications 21.5, pp20-26.
29) Chellaprabha, B., and S. Chenthur Pandian, (2012) "A Multipath Energy Efficient Congestion Control Scheme for
Wireless Sensor Network", Journal of Computer Science 8, 2012.
30) Cherian, Mary, and T. R. Nair, (2012) "Multipath routing with novel packet scheduling approach in wireless
sensor networks", arXiv preprint arXiv:1206.0420.
31) Liu, Wenjun, Shukui Zhang, and Jianxi Fan, (2012) "A Diagnosis-Based Clustering and Multipath Routing
Protocol for Wireless Sensor Networks", International Journal of Distributed Sensor Networks.
32) Vidhyapriya, R., and P. T. Vanathi, (2007) "Energy efficient adaptive multipath routing for wireless sensor
networks", IAENG International Journal of Computer Science 34.1, pp56-64.
33) Ramadan, Rabie A., (2009) "Agent based multipath routing in wireless sensor networks." Intelligent Agents, 2009.
IA'09. IEEE Symposium on.
34) Sutagundar, A. V., and S. S. Manvi, (2012) "Wheel Based Event Triggered Data Aggregation and Routing in
Wireless Sensor Networks: Agent Based Approach", Wireless Personal Communications, pp1-27.
35) Herbert Taub, Donald L. Schilling, (1986) “Principles of communication systems”, McGraw-Hill.
36) Miao, Guowang, Nageen Himayat, and Geoffrey Ye Li., (2010) "Energy-efficient link adaptation in frequency-
selective channels", Communications, IEEE Transactions on 58.2 pp545-554.
37) Sutagundar, A. V., and S. S. Manvi, (2013) "Location aware event driven multipath routing in Wireless Sensor
Networks: Agent based approach", Egyptian Informatics Journal.
38) Heinzelman, Wendi Rabiner, Anantha Chandrakasan, and Hari Balakrishnan, (2000) "Energy-efficient
communication protocol for wireless microsensor networks", System Sciences, 2000. Proceedings of the 33rd
Annual Hawaii International Conference on. IEEE.
Authors
V. S. BENNUR completed master degree in digital communication, Visvesvaraya Technological
University, Belgaum, Karnataka. Presently working as Assistant Professor for Department of
Electronics and Communication Engineering, East Point Woman Engineering College, Bangalore.
She is presently perusing her Ph.D in Visvesvaraya Technological University, Belgaum. Her area
of interest includes network security, Wireless Sensor Networks, image processing, error control
coding, etc., She published 5 national/international conference paper.
S. S. SHIRABUR obtained his bachelor degree in Electronics and Communication Engineering and
Master degree in Digital Communication from Visveswaraya Technological University, Belgaum,
Karnataka. He is currently working as an Assistant Professor for Department of Electronics and
Communication Engineering, Veerappa Nisty Engineering College Shorapur, Karnataka, India. His
area of interest includes Wireless networks, Multimedia Networks, Computer communication
networks and Image Processing.
A. V. SUTAGUNDAR completed his Ph.D. from Visvesvaraya Technological University,
Belgaum, Karnataka. Presently he is serving as an Associate Professor of Department of Electronics
and Communication Engineering Bagalkot, Karnataka. His areas of interest include Signal and
system, Digital Signal Processing, Digital Image Processing, Multimedia Networks, Computer
Communication networks, Wireless networks, Mobile ad-hoc networks, Agent technology. He has
published 26 papers in referred National/International Conferences and 11 papers in international
journals.

Multiagent based multipath routing in wireless sensor networks

More Related Content

What's hot (17)

Viewers also liked (18)

Similar to Multiagent based multipath routing in wireless sensor networks (20)

Recently uploaded (20)

Multiagent based multipath routing in wireless sensor networks