50120140503002

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-6367(Print),

ISSN 0976 - 6375(Online), Volume 5, Issue 3, March (2014), pp. 09-22 © IAEME

9

3-D LOCALIZATION IN WIRELESS SENSOR NETWORK

Mr. Om Mehta1

IITE, Indus University, Ahmadabad

Prof. Gaurang Raval2

Institute of Technology, Nirma University

ABSTRACT

Wireless sensor networks have a great impact on long time monitoring applications

(environment monitoring, security surveillance, habitat monitoring etc.) but its potential is et to be

discovered, where it can be deployed in time critical situations when disaster happens. There is a

significant gap between existing applications of sensor networks and the requirement of applications

supporting rescue operations that involves the catastrophe of human lives. As we are dealing with the

human lives here, we can’t just rely on the localization schemes that depend upon the connectivity

information (range- free) algorithms only. Further, rescue operations are carried out in highly noisy

environments, so distance based (range-based) localization algorithms generate high error in distance

measurements. An efficient algorithm is needed that can measure the location of the sensor nodes

near to the living being or being attached to them in 3-D space with a high accuracy. To achieve

such kind of accuracy a combination of both the strategies is required.

Keywords: Wireless Sensor Networks, Localization, Range Free, Range Based approach

I. INTRODUCTION

When we face an unfortunate situation such as an earth- quake, hurricane, or similar

disasters, wireless sensor nodes prove to be very useful in search and rescue operations. The wireless

sensor network can be formed on the fly without any previous existing infrastructure. Wireless

sensor networks are dynamically formed over the varying topologies. Wireless sensor networks can

assist in conducting the rescue operations and can provide search in timely manner. In general,

disasters leave the situation worse without power and disruption of communication capabilities

destroying the infrastructures [3]. A sensor network mainly consists of anchor nodes as well as the

other nodes sensing the data. The anchor nodes are location aware sensor nodes which obtain its

location information through some external methods (GPS, TOA [1, 5]). The un- localized nodes

hear the beacons which are broadcasted by the anchor nodes. Determining the location of these un-

localized sensor nodes with respect to the anchor nodes is called localization [2]. The localization

INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING &

TECHNOLOGY (IJCET)

ISSN 0976 – 6367(Print)

ISSN 0976 – 6375(Online)

Volume 5, Issue 3, March (2014), pp. 09-22

© IAEME: www.iaeme.com/ijcet.asp

Journal Impact Factor (2014): 8.5328 (Calculated by GISI)

www.jifactor.com

IJCET

© I A E M E



10

algorithms generally possess wave-spreading like characteristics in which the intermediate sensor

nodes become location aware and act as a new reference anchor node for the un-localized

neighboring nodes.

All the strategies that are employed for the localization in 2-D spaces are violated in 3-D

spaces [9, 4].2-D spaces cannot be directly extended to 3-D just by increasing one parameter. For

example, the 2-D localization completely fails in determining the depth of the river bed and other

similar sensor network scenarios where the height comes into play. There are several well known

problems that can be easily solved modeling the sensor network as 2-D but are very complex

(sometimes NP- Hard) when modeled as 3-D. Moreover, the main triangulation approach that is

employed for the 2-D space doesn’t fit for the 3-D [7].

The information received from the remote sensors that are several hops far is meaningless

until and unless the location from where the data is received is known [3]. For large scale

application scenarios where the geographical data is to be known for coarse parameters such as

the temperature of an area, the data is aggregated for all the sensor nodes that are located

nearby. Further, the location information is important for many location aware sensor network

protocols. The positional information is used to partition the network into several clusters for

hierarchical routing. For efficient routing the protocols require the information of the location of the

sensor nodes. The sensors employed near the disaster affected sites might contain the vital

parameters of the living being trapped beneath several layers of collapsed surfaces. Based on the size

of the area affected by catastrophe the data can be directed to the rescue team in a multi-hop way.

The data so received by the rescue team can be processed and a map can be generated displaying

the optimized path to send the help service in minimum possible time.

II. RELATED RESEARCH WORK

The wireless sensor network is an open research field. All the components of this network

are still in developing phase. My different localization algorithms have been proposed until now. As

the sensor networks are application specific it is quite hard to generalize the algorithm approaches

for localizations that suits to all the different scenarios. All the localization algorithms have been

broadly classified into two major cate- gories: range based and range free. Both the strategies

employ totally different way of approaches. The range-based method relies on the distance

measurements through different received signal strength strategies. Whereas, the range-free method

relies on the connectivity information only i.e. the nodes can listen to the beacon from the sensor

nodes. The wireless sensor network consists of localized anchor nodes and un-localized sensor

nodes. The range free method uses the connectivity information and bound the nodes location to the

common over- lapped (intersected) area. Different algorithms use different methods to calculate the

nodes location information within the bounded region with respect to the anchor nodes.

The range-based method uses the sophisticated hardware, radio signals to estimate the

distance between the receiver and the transmitter antennas. Moreover, the range-free approach

reuses the wireless communication radio signals to determine the connectivity between neighboring

sensors and requires no extra hardware support. The range-free methods are used where the precise

location of the sensor nodes is not required and the coarse position estimation up to some level is

tolerable.

III. RANGE-BASED LOCALIZATION SCHEMES:

A. Received Signal Strength (RSS):

RSS is defined as the power measured by a received signal strength indicator (RSSI) by the

receiver. Often, RSS is equivalently reported as measured power [5], i.e., the squared magnitude of



11

the radio signal strength [2, 8]. We can consider the RSS of low frequency, RF, or other signals.

Wireless sensors nodes communicate with neighboring sensors, so the signal strength of the radio

can be calculated by each receiver during normal data communication without requiring additional

bandwidth or energy requirements. RSS measurements are relatively can be easily embedded in the

motes, thought of their expensiveness they are most generally used method for distance calculation

[7].The power of the receiving radio signals can be calculated by measuring the voltage[2]:

P = −51.3 × RSSI − 49.2 (1)

Noise is the standard deviation σ of all RSS values that may be observed at a particular

distance in a given environment. Even with adjacent stationary sensor nodes, RSS will vary to

some degree due to noise. The noise affecting the RSS is not constant; it varies with time as well

as environmental conditions. Finally, radio signals can vary significantly in both transmission

strength and receptivity, particularly in cheap, low-power radio devices [4, 5]. The RSS directly

varies with the inverse power of distance. The attenuation rate is the rate σ at which radio signal

strength decreases over distance:

RSS α d−α (2)

Generally, if α =2σ then radio signal strength drops by 3dB every time distance doubles.

Generally 2σ for highly noisy environments. This linear attenuation rate means that the difference

between the radio signal strength varies according to the predefined rate for example; difference

between 1meter and 2meter is similar to the difference between 10meter and 20meter i.e. exactly

3dB. Moreover, a constant level of noise can result in increasing error rates when signal strength is

used to estimate distance.

The received signal strength (RSS) method uses the relationship of radio frequency signal

as a function of distance. Now, from the relationship mathematical propagation model also be

derived. studies shows that of the Radio Frequency signal propagation characteristics [7], the

main disadvantages of the methods are that the propagation characteristics of radio signals

could be vary with changes in the surrounding environment temperatures, humidity, pressures, areas

such as: rural, urban. It also suffers multipath effect (weather changes, urban, rural and indoor,

outdoor settings) until one direct line of site the RSS detection is somewhat difficult.

B. Time Based Methods (TOA & TDoA): Time of arrival (TOA) and Time Difference of Arrival (TDoA) calculates distances which

based on the propagation time and that propagation time can directly translated into distance, based

on the known signal propagation speed. These methods can be applied to many different signals,

such as Randio Frequency, acoustic, infrared and ultrasound [5]. They relies on complex hardwares

that is expensive and energy consuming; making it less suitable for sensor network where the

scalability is high of the lifetime of the network is expected to be more.

TDOA is a particular case of TOA approaches are utilized when the source clock time

is known to us. This strategy is analytical in nature and uses the TDOA between pairs of sensor

nodes. The resulting equations are nonlinear but can be reduced to linear in the solution process. The

approach is based on signaling sources sending low frequency at known, locations in and around

the sensor field. Each source broadcasts in a round- robin fashion and the time-of-arrival is

recorded by each sensor node. The differences in the time are calculated and the co-efficient

equations are themselves can then be communicated back to the sensor nodes for calculations.

While radio frequency based localization is promising; there are several reasons for low frequency

signaling techniques to be more desirable. Firstly, the relatively low speed wave results in less

costly hardware as compared to high-bandwidth radio frequency hardware. The low frequency wave

also means that the system will be less sensitive to errors in time-of- flight measurements,

synchronization along with other timing errors. However, low frequency techniques also have

somete disadvantages over radio frequency techniques. Firstly, low frequency waves are highly



12

deviated by the environmental conditions such as temperature, humidity and pressure resulting in

uneven wave speeds over the area of interest. Secondly, low frequency waves suffer greater

attenuations through edges of buildings and other physical objects rendering it nearly inefficient

[6].

IV. RANGE-FREE LOCALIZATION SCHEMES

A. Centroid Algorithm:

Anchors send their location information to neighbors that keep an account of all received

beacons. Using the averaged out information, simple centroid model is applied for estimation the

listening nodes location. This protocol is referred as the Centroid algorithm.

The following are the steps of the localization:

• Beacon node broadcasts their position.

• Sensor node listens for beacons from anchors nodes and if they are able to hear the beacons

this means they are under the ranging area covered by these anchor nodes.

• Sensor node computes its position by averaging out all the beacon node locations.

In [6] a range-free, proximity-based, a localization algorithm is proposed containing

location information (xi , yi ) that uses anchor beacons, to estimate sensor node position. After

receiving these beacons from anchor nodes, a sensor node estimates its location using the following

formula for 2-D space [3]:

Estimate its location. The accuracy of the DV-hop method increases as the scalability of the

network increases. The averaged value with more number of anchor nodes produces less error than

the estimated value with less no of anchor nodes with a high deviation from the actual location.

(xest , yest ) = (x1 + x2 + ... + xn , y1 + y2 + ... + yn ) (3)

N N

The advantage of this Centroid localization scheme is its simplicity of calculation and

simplicity in implementing it.

Fig.1. Centroid estimation

B. DV-HOP: DV-HOP assumes that a network consisting of identical sensing nodes and beacon nodes

[5]. Instead of going for the single hop broadcast, anchors flood their locations all over the sensor

network. As the information propagates over each hop an increment counter each time increments

the hop count value. Senor nodes calculate their position based on the received beacon locations,

average distance per hop, the hop count from the corresponding anchor; the working of this strategy



13

is quite similar to existing distance vector routing. One anchor node broadcasts a beacon to all

over the network containing the beacon location with a hop count value initially set to one.

The receiving sensor nodes maintain a minimum counter value corresponding to each

anchor. Every time it receives a beacon it increments the hop count value and ignores those

anchors with higher hop-count values. All the way propagating through this mechanism, all sensor

nodes in the network (including other beacons) get the shortest distance, in hops. Beacons perform

this task by obtaining position and hop count information for all other beacons inside the network.

The average one hop distance is then calculated by beacon i using the following formula:

In the above formula (xm, ym) is the location of anchor m, and hm is the calculated distance

in hops, from anchor m to beacon b. Once calculated, beacons spread the estimated size of the hop

information out to the nearby sensor nodes. Once a sensor node calculates the distance estimation to

more than 3 anchors in the plane, it uses triangulation (multilateration) method to estimate its

location. The accuracy of the DV-hop method increases as the scalability of the network increases.

The averaged value with more number of anchor nodes produces less error than the estimated value

with less no of anchor nodes with a high deviation from the actual location.

V. P ROPOSED ALGORITHM

The following steps are performed for calculation of 3-D location:

1) The whole area is divided into grids. The interspacing between the grids is less than the range of

the anchor nodes.

2) A beacon node covers a block of cubes that are within its radio range. The cube size is such

that it covers the largest area in the sphere formed by the radio range.

3) The distance between the un-localized sensor node and the beacon nodes is calculated using

different RSS techniques.

4) A list is maintained by each un-localized node that contains the number of beacon nodes and the

distance from each beacon node.

5) The common intersection area gives the bounded values (max, min).The bounded values

specifies the maximum and minimum values of co-ordinates. The nodes position should be

within this rectangular region.

6) The common intersection area gives the bounded β values (max, min).The bounded values

specifies the maximum and minimum values of co-ordinates. The nodes position should be

within this rectangular region.

7) Check whether the beacon nodes (≥ 3) are forming a plane i.e. they are not on same line.

8) Calculate the location (x, y, z) of un-localized nodes taking the distances into account.

9) Check the whether the calculated values within bounding region. All the values outside the

region are discarded.

10) Change the status of newly localized node to new beacon nodes for further calculation of other

nodes.

A. Range Free Approach Instead of considering a spherical area around a sensor node we consider the sensor

node covers a cubical area. All the nodes within the cubical area can listen to anchor node



14

other outside it are not considered to be covered by this anchor node. The position of the node

is guaranteed to lie in the intersection of the bounding cubes corresponding to all the beacons

within the radio range. This set itself is a rectangular (sometimes square) cube, whose minimum and

maximum values are computed by iterating the process upon all the beacons heard: at each stage,

the new boundary values are compared to the previous ones, and the smallest set is

In the above formula (xm, ym) is the location of anchor m, and hm is the calculated

distance in hops, from anchor m to beacon b. Once calculated, beacons spread the estimated size of

the hop information out to the nearby sensor nodes. Once a sensor node calculates the distance

estimation to more than 3 anchors in the plane, it uses triangulation (multilateration) method to

kept. Cube is the only regular polyhedron that tessellates a 3-D space. Each cube representing the

bounding area can be represented by the min max values of its coordinates. The radio range

covered by the beacon node P(x, y, z) varies between [xmin , ymin , zmin ] and [xmax ,

ymax , zmax ] represents the bounded range β of a node. r average radio range of the beacon.

[xmin , ymin , zmin ] ≤ β ≤ [xmax , ymax , zmax ]

Fig.2. Intersection of bounding boxes

Fig.3.Bounded Value of Sensor Nodes

The intersection of the two cubical areas is done to reduce the area where the un-localized

node is likely to be. As, more number of cubical areas start to merge the common area of

intersection starts to reduce. The new rectangular (sometimes square) area that emerges after

multiple intersections of beacon nodes represents the bounded area.

For example intersection of two bounding cubes;

βnew = β1 I β2 (5)



15

Similarly, intersection of multiple boundaries can be done to lower the number of positions

where a sensor node can be present.

βnew = β1 I β2 ........I βn

The method falls under the category of range-free as it only uses the connectivity

information i.e. whether it is able to send receive the beacon nodes from the respective beacon

nodes.

B. Range Based Approach The method uses the distance information between the un- localized node and the beacon

nodes. All the beacon nodes are considered to lie on same plane, w.r.t of these beacon nodes the

nodes localization is determined. A set of three anchor nodes are selected at a time. These three

anchor nodes form a equation of a plane. The standard equation of a plane in 3 dimensional spaces

is:

Qx + Ry + Sz + D = 0 (7)

Here in equation (19), Q, R, S correspond to the anchor nodes D corresponds to the sensor

node with unknown location information.

The normal to the plane is the vector (Q, R, S). Given three points in space Q(x1, y1, z1), S(x2, y2,

z2), S(x3, y3, z3) the equation of the plane through these points is given by the following

determinants.

Expanding the above gives:

Q = y1 (z2 − z3) + y2 (z3 − z1) + y3 (z1 − z2) (8)

R = z1 (x2 − x3) + z2 (x3 − x1) + z3 (x1 − x2) (9)

S = x1 (y2 − y3) + x2 (y3 − y1) + x3 (y1 − y2) (10)

D = −[x1 (y2 z3 − y3 z2) + x2 (y3 z1 − y1 z3) + x3 (y1 z2 − y2 z1)] (11)

The sign of, s = Qx + Ry + Sz + D determines which side the point (x,y,z) lies with

respect to the plane. If s > 0 then the point lies on the same side as the normal (Q, R, S). If s <

0 then it lies on the opposite side, if s = 0 then the point (x,y,z) lies on the plane. pi is the

distance between the point P and the beacon nodes . The distances are measured by using

different received signal strength (RSS) strategies. At least three beacon nodes are required to

calculate the location of unknown sensor node. For a plane Qx+Ry+Sz+D=0 and a point P

(x1 , y1 , z1 ) not necessarily lying on the plane, the shortest distance from P to the plane is:



16

Fig.4. Distance of point P from a plane

It follows that if Distance = 0 in equation then P lies in the same plane.

However, in real life environmental conditions where there is interferencing noise the

measuring and calculating errors might change the distance a little bit. As a result P is considered

lying within the same plane if it satisfies the constraint:

−ζ < Distance < ζ (13)

Where ζ is minute deviation caused due to errors. At first it is determined whether there is node

formed: (x − xi )+ (y − yi )+ (z − zi )= p2 (14)

Fig.5. unknown coordinates of P

Solving the above equation we can get multiple values of z because of quadratic equations.By

using proper bounded cubes the proper values of (x,y,z) is being selected so that it satisfies

β = β1 | β2 | β3 (15)

The β cube is calculated using the intersected cubical Rectangle of beacon nodes. Pєβ

If they are not lying on the same plane then multiple beacon values giving rise to multiple equation

are used for finding the values of p(x,y,z) and minimize the error:

(x − xn )+ (y − yn )+ (z − zn )− d2 = 0 (16)

Where (xn , yn , zn ) corresponds to the location of the beacon nodes.



17

Solving the above equation approximated distance can be calculated:

p = (αT α)−1 αT k (17)

Further, we bound the multiple values of x,y,z by selecting only those values which are

within the bounding region and rejecting the one which are outside it. The bounding region is where

we have the constrained area of x, y, z. Now substituting the value z in the other equation we can get

the proper value of (x,y,z).MICA2 motes can be also used to check the feasibility of proposed

Algorithm.

VI. SIMULATION

The algorithm proposed here, as all the geometric approaches of the localization problem,

requires a large number of sensor nodes and location aware beacons. As we only have a few sensor

nodes, this could not be practically implemented on our MICA2 motes. All the simulations related

to thesis is done in MATLAB.

MATLAB [18] simulator is selected as it easily handles the matrix calculations very

effectively all the stress of the work can be given to the actual work rather than going on

programming details.

We have compared the parameters with a standard Center of Gravity (COG) method. The

simulation is performed by taking 1/3rd of the total nodes as beacon nodes. The total time

calculated here is the combined simulation time of both the proposed one and the COG method.

Giving, an estimation of how much time this proposed algorithm will take to execute.

A. Performance Analysis The percentage of success of the algorithm proposed in this thesis is given by:

di deviation between the actual location of nodes and the estimated location. N total number

of sensor nodes in the network (considering the first layer. contains only beacon nodes). n number

of nodes per layer B. Error Analysis The error is calculated using:

derror = √(xa − xc )2 + (ya − yc )2 + (za − zc )2 (19)



18

where,

xa , ya , za - are actual sensor node location

xc , yc , zc - are calculated sensor node location.

VII. RESULT

A. Effect of Scalability on the Performance: Here, in graph we can clearly see that as the number of nodes is increased the performance of

the algorithm improves by 10 % - 30 %. Whereas, in COG method the performance improves very

less by 5 %-15 %. For 20 anchor nodes the performance of success is around 60 %, as the no of

anchor nodes are increased the success rate rises above 90 %.It supports the fact that as the

number of beacon nodes in the network increases the calculated location values becomes more

close to the original location.

Fig.6. Anchor nodes vs Percentage Beacons

Here, table shows the sensor network nodes that are used to generate the simulated output.

The beacon nodes are distributed throughout the first layer. All the other sensor nodes are randomly

distributed to respective layers with equal node densities. The uneven horizontal mesh corresponds

to the surface beneath.



19

Fig7.Simulation done by 50 anchor nodes, total number of 150 nodes

Fig.8. Simulation done by 500 anchor nodes, total number of 1500 nodes

B. Effect of Noise: The noise here is considered to be Gaussian Noise with mean µ = 0 and standard

deviation σ.Here, we can clearly see that the noise has direct impact on estimation accuracy. As

the noise value increases the success rate of the proposed scheme varies between 55% - 95%.

Though the success rate drops but still it gives better results than other methods.

Fig.9.Noise factor vs Percentage success



20

Fig.10. Simulated output taking 150 nodes standard deviation of 0.005

Fig.11.Simulated output taking 150 nodes standard deviation of 0.025

In COG the noise has very less impact because of the nature of COG estimation of value. As

COG method considers the mean of all the locations so a localization error tends to deviate the

calculated values very less.

C. Effect of Radio Range

The simulation is carried out taking the inter layer separation 3m. As the proposed

utilizes both the range-free and range-based strategies there is not much performance change in it.

Further, in case of COG method initially when the radio range is very less the performance of

COG is quite high, intermediate radio ranges the success rate varies between 25%-50%, as the

range of the anchor nodes increases one layer of sensor nodes are affected by multiple layers of

beacon nodes above it giving it a increasing success rate.



21

Fig.12. Simulation done taking 150 nodes with radio range 5m

The simulation time taken increases irrespective of the number of sensor nodes as for single layer

information of location from multiple layers of beacon nodes are processed.

VIII. CONCLUSION

The literature survey work has shown that an optimal algorithm has not been

defined yet, that employ both the strategies (range-free range-based) , and thus the development

of a new algorithm has to be founded on the specificities of the situations, taking into account the

size of the network, as well as the deployment methods and the expected results. Localization

algorithms should be designed to achieve low variance as well as low bias as far as possible; at

the same time, they need to be scalable to very large network sizes without dramatically increasing

energy consumption or computational requirements. We have proposed and demonstrated an

algorithm that successfully localizes nodes in a sensor network with noisy distance

measurements. The equations for the proposed algorithm were carried out in MATLAB.

Simulations showed the relationship between noise and ability of a network to localize itself, at

highly noisy environments. The performance stays above 50% i.e. half of the available nodes can be

localized with good approximation.

We also have depicted the effect of scalability on the performance of the algorithm. Results

show that as the scalability of the network increases with the number of beacon nodes; the

performance of the algorithm goes high above 90%. The granularity of the areas estimated may be

easily adjusted by changing the system parameters which makes the proposed algorithm flexible.

REFERENCES

[1] Vibha Yadav, Manas Kumar Mishra, A.K.Sngh and M.M.Gore, ”LOCALIZATION SCHEME

FOR THREE DIMENSIONAL WIRE- LESS SENSOR NETWORKS USING GPS ENABLED

MOBILE SEN- SOR NODES” International Journal of Next-Generation Networks

(IJNGN),Vol.1, No.1, December 2009.

[2] Rong Peng and Mihail L. Sichitiu,”Angle of Arrival Localization for Wireless Sensor

Networks,Department of Electrical and Computer Engieering North Carolina State University

Raleigh, NC 27695.



22

[3] AZZEDINE BOUKERCHE, EDUARDO F. NAKAMURA, ANTONIO A.F. LOUREIRO.

LOCALIZATION SYSTEMS FOR WIRELESS SENSOR Networks, IEEE Wireless

Communications, December 2007

[4] M. Keshtgary, M. Fasihy, and Z. Ronaghi LOCALIZATION SYSTEMS FOR WIRELESS

SENSOR Networks, IEEE Wireless Communications, January 2011.

[5] S. M. Nazrul Alam, Zygmunt J. Haas, Topology Control and Network Lifetime in Three-

Dimensional Wireless Sensor Networks, 2006

[6] Holgar Karl and Andreas Willig, Protocols and Architectures for Wire- less Sensor Networks.

[7] Amitangshu Pal, Localization Algorithms in Wireless Sensor Networks: Current Approaches

and Future Challenges, 2010.

[8] Kamin Whitehouse, Chris Karlof, David E. Culler, A practical evaluation of radio signal

strength for ranging-based localization. Mobile Computing and Communications Review

11(1): 41-52, 2007.

[9] Schmidt, Peter L., Williams, Stephen M. And Frampton, Kenneth D, Acoustic self-

localization of a wireless sensor network, 2005.

[10] Chong Han, Lijuan Sun Fu Xiao, Jian Guo, and Ruchuan Wang ,”A Camera Nodes

Correlation Model Based on 3D Sensing in Wireless Multimedia Sensor Networks”,2012.

[11] Hamid Shokrzadeh,P.Saadatmndi, M. H.Goodarzi,”Rumor Routing by Appointment in Center

of Gravity in Wireless Sensor Networks”,2011. [12] I an F.Akyildiz,” Wireless Sensor

Networks”.

[13] Fusion David J. Scherba and Peter Bajcsy, Depth Map Calibration by

Stereo and Wireless Sensor Network July,2007.

[14] Jian Ma, Qianbin Chen, Dian Zhang and Lionel M. Ni, An Empirical study of Radio Signal

Strength in Sensor Networks using MICA2 Nodes, march 2006.

[15] Mohammad Ilyas and Salahuddin Qazi, Adhoc Wireless Networks Challenges Opportunities

[16] Jianlin Liang, Jun Shao and Ying Xu, Constrained 3-D Localization in Sensor Networks

[17] Wang Chen, Xiao Li, Jin Rong, Sensor Localization in an Obstructed Environment. DCOSS

2005: 49-62

[18] Sameera Poduri, Sundeep Pattem, Bhaskar Krishnamachar, Sensor Net- work Configuration

and the Curse of Dimensionality

[19] Jonathan Bachrach and Christopher Taylor, Localization in Sensor Networks, 2005.

[20] Niculescu and Badri Nath, Ad Hoc Positioning System (APS) Using AOA Dragos, 2009.

[21] Wategaonkar D.N and Deshpande V.S., “On Improvement of Performance For Transport

Protocol Using Sectoring Scheme In WSN”, International Journal of Computer Engineering &

Technology (IJCET), Volume 4, Issue 4, 2013, pp. 275 - 281, ISSN Print: 0976 – 6367, ISSN

Online: 0976 – 6375.

[22] Bharathi M A, Vijaya Kumar B P , Manjaiah D.H, “Power Efficient Data Aggregation Based

on Swarm Intelligence and Game Theoretic Approach In Wireless Sensor Network”,

International Journal of Computer Engineering & Technology (IJCET), Volume 3, Issue 3,

2012, pp. 184 - 199, ISSN Print: 0976 – 6367, ISSN Online: 0976 – 6375.

[23] Mr.Yogesh V Patil, Mr. Pratik Gite and Mr.Sanjay Thakur, “Automatic Cluster Formation and

Assigning Address For Wireless Sensor Network”, International Journal of Computer

Engineering & Technology (IJCET), Volume 4, Issue 4, 2013, pp. 116 - 121, ISSN Print: 0976

– 6367, ISSN Online: 0976 – 6375.